#  @AiBattle_ AiBattle AiBattle posts on X about $googl, prompt, in the, the new the most. They currently have [-----] followers and [---] posts still getting attention that total [------] engagements in the last [--] hours. ### Engagements: [------] [#](/creator/twitter::1367154052409212928/interactions)  - [--] Week [-------] +47% - [--] Month [-------] -0.30% - [--] Months [---------] +201% ### Mentions: [--] [#](/creator/twitter::1367154052409212928/posts_active)  ### Followers: [-----] [#](/creator/twitter::1367154052409212928/followers)  - [--] Week [-----] +3.80% - [--] Month [-----] +26% - [--] Months [-----] +92% ### CreatorRank: [-------] [#](/creator/twitter::1367154052409212928/influencer_rank)  ### Social Influence **Social category influence** [technology brands](/list/technology-brands) 35.9% [stocks](/list/stocks) 18.8% [gaming](/list/gaming) 2.56% [social networks](/list/social-networks) 1.71% [finance](/list/finance) 1.71% [events](/list/events) 1.71% [countries](/list/countries) 1.71% [celebrities](/list/celebrities) 0.85% [products](/list/products) 0.85% **Social topic influence** [$googl](/topic/$googl) 17.09%, [prompt](/topic/prompt) 10.26%, [in the](/topic/in-the) 7.69%, [the new](/topic/the-new) 7.69%, [ai](/topic/ai) 6.84%, [o3](/topic/o3) 5.98%, [open ai](/topic/open-ai) 5.13%, [bytedance](/topic/bytedance) #66, [context window](/topic/context-window) #1, [anthropic](/topic/anthropic) 4.27% **Top accounts mentioned or mentioned by** [@hibara_ai_lover](/creator/undefined) [@aquiffoo](/creator/undefined) [@chetaslua](/creator/undefined) [@v0](/creator/undefined) [@scaling01](/creator/undefined) [@razroochief](/creator/undefined) [@noname890098](/creator/undefined) [@bdsqlsz](/creator/undefined) [@tonitrades_](/creator/undefined) [@aero96193997](/creator/undefined) [@aoligei15](/creator/undefined) [@slartibart_](/creator/undefined) [@girish_lelouch](/creator/undefined) [@drbeavisai](/creator/undefined) [@gauravdhiman_ai](/creator/undefined) [@apathium65906](/creator/undefined) [@xiaotiaowang](/creator/undefined) [@streyai](/creator/undefined) [@wokoma_festus](/creator/undefined) [@__satoshissss__](/creator/undefined) **Top assets mentioned** [Alphabet Inc Class A (GOOGL)](/topic/$googl) [Microsoft Corp. (MSFT)](/topic/microsoft) [Trex Company, Inc. (TREX)](/topic/$trex) ### Top Social Posts Top posts by engagements in the last [--] hours "New Grok image model is being tested on LmArena under the name "Sumo"" [X Link](https://x.com/AiBattle_/status/2005970763019608196) 2025-12-30T11:54Z [----] followers, 25.3K engagements "ByteDance has been testing its new Doubao model for a week now in Kilo Code under the name "Giga-Potato" Description from Kilo Code: "In our internal benchmarks it has outperformed nearly every open-weight model weve tested on long-context coding tasks - Context Window: 256k Tokens - Max Output: 32k Tokens - Strict Adherence: The model is showing exceptional discipline in following system prompts making it ideal for enterprise environments with strict linting and style guidelines" https://twitter.com/i/web/status/2014361796279181388 https://twitter.com/i/web/status/2014361796279181388" [X Link](https://x.com/AiBattle_/status/2014361796279181388) 2026-01-22T15:37Z [----] followers, 15.4K engagements "Kimi K2.5 is live on the Kimi website The model looks really promising on zero-shot coding prompts so far. Well see how it translates to agentic coding tasks but so far Im really excited" [X Link](https://x.com/AiBattle_/status/2015902394312253564) 2026-01-26T21:39Z [----] followers, 50.2K engagements "Kimi K2.5 scores 46.8% on SimpleBench DeepSeek V3.2 Speciale remains the open-weights model with the highest score on this benchmark at 52.6%" [X Link](https://x.com/AiBattle_/status/2016504175123677469) 2026-01-28T13:30Z [----] followers, 15.8K engagements "New Claude model update(s) are coming The upcoming "Fennec" model (Sonnet update) seems to be better than Opus [---] according to tests from @chetaslua Big week for Anthropic fans coming up😉 (Or perhaps just anyone who uses AI to code) Big week for Anthropic fans coming up😉 (Or perhaps just anyone who uses AI to code)" [X Link](https://x.com/AiBattle_/status/2017619997338538103) 2026-01-31T15:24Z [----] followers, 224.7K engagements "Newly released Stepfun model "Step-3.5-Flash" beats DeepSeek v3.2 on several benchmarks while having far fewer parameters Step-3.5-Flash: 196B total / 11B active Parameters DeepSeek v3.2: 671B total / 37B active Parameters This week / month will likely have some of the most impactful model releases in a while from both U.S. and Chinese labs Exciting days ahead https://twitter.com/i/web/status/2018143041840697520 https://twitter.com/i/web/status/2018143041840697520" [X Link](https://x.com/AiBattle_/status/2018143041840697520) 2026-02-02T02:02Z [----] followers, 40.8K engagements "@v0 just deleted the Sonnet [--] tweet @scaling01 was right it was just engagement baiting" [X Link](https://x.com/AiBattle_/status/2019061374194721131) 2026-02-04T14:51Z [----] followers, [----] engagements "Opus [---] has been found in the Perplexity API "Anthropic's most advanced model" 🚨NEW: Claude Opus [---] & Claude Opus [---] Thinking are now live on Perplexity's APIs Looks like we're getting it today and Sonnet [--] later https://t.co/pL3m2yOyUd https://t.co/glaky9eAAP 🚨NEW: Claude Opus [---] & Claude Opus [---] Thinking are now live on Perplexity's APIs Looks like we're getting it today and Sonnet [--] later https://t.co/pL3m2yOyUd https://t.co/glaky9eAAP" [X Link](https://x.com/AiBattle_/status/2019385374842495185) 2026-02-05T12:19Z [----] followers, 29.5K engagements "Claude Opus [---] has the same pricing as Claude Opus 4.5" [X Link](https://x.com/AiBattle_/status/2019466880667332851) 2026-02-05T17:43Z [----] followers, 23.3K engagements "Benchmark scores for GPT-5.3-Codex (xhigh)" [X Link](https://x.com/AiBattle_/status/2019473343334731926) 2026-02-05T18:08Z [----] followers, [----] engagements "Claude Opus [---] scores 67.6% on Simple-Bench" [X Link](https://x.com/AiBattle_/status/2019691886244688191) 2026-02-06T08:37Z [----] followers, [----] engagements "The Information reports new models from ByteDance and Alibaba next month: - ByteDance will launch three new AI models next month - Alibaba will release their next-gen model next month Separately it's likely we're getting the next-gen DeepSeek model next month too Chinese New Year is going to be crazy ByteDance and Alibaba are set to launch new AI models (The Information). ByteDance plans to unveil three new AI models next month and Alibaba is also expected to introduce its next-generation AI model next month. ByteDance and Alibaba are set to launch new AI models (The Information). ByteDance" [X Link](https://x.com/AiBattle_/status/2016927443051798593) 2026-01-29T17:32Z [----] followers, 28.3K engagements "Potential new Qwen and ByteDance Seed models are being tested on the Arena The Karp-001 and Karp-002 models claim to be Qwen-3.5 models The Pisces-llm-0206a and Pisces-llm-0206b models claim to be ByteDance models" [X Link](https://x.com/AiBattle_/status/2020101939887829367) 2026-02-07T11:46Z [----] followers, 45.8K engagements "GLM-5 scores higher than Gemini [--] Pro on the Artificial Analysis Intelligence Index GLM-5 is now the open-weight model with the highest score" [X Link](https://x.com/AiBattle_/status/2021662669955379205) 2026-02-11T19:08Z [----] followers, 45.5K engagements "Opus [---] performs worse than Opus [---] on SWE-Bench" [X Link](https://x.com/AiBattle_/status/2019467939536093630) 2026-02-05T17:47Z [----] followers, 84.8K engagements "Claude Opus [---] ARC-AGI scores are live Opus [---] (120k High) has the highest score: ARC-AGI-1: 94.00% ARC-AGI-2: 69.20%" [X Link](https://x.com/AiBattle_/status/2019484234226774199) 2026-02-05T18:52Z [----] followers, [----] engagements "It claims to be a Claude model. Anthropic doesnt do stealth model releases as far as I know so its likely from a Chinese lab It has a 200k context window the same as the GLM [---] and [---] models so it might be a new GLM model The GLM models are also known for sometimes identifying themselves as Claude models https://twitter.com/i/web/status/2019836771484184862 https://twitter.com/i/web/status/2019836771484184862" [X Link](https://x.com/AiBattle_/status/2019836771484184862) 2026-02-06T18:13Z [----] followers, [----] engagements "A new stealth model "Pony-Alpha" is being tested on OpenRouter" [X Link](https://x.com/AiBattle_/status/2019830954513277430) 2026-02-06T17:50Z [----] followers, 13.1K engagements "Pelican SVG looking good" [X Link](https://x.com/AiBattle_/status/2019831714571514055) 2026-02-06T17:53Z [----] followers, [----] engagements "Qwen [---] has been spotted on GitHub - Qwen3.5-9B-Instruct - Qwen3.5-35B-A3B-Instruct The [--] currently available models on the Arena "Karp-001" and "Karp-002" could possibly be the small Qwen-3.5 models Potential new Qwen and ByteDance Seed models are being tested on the Arena The Karp-001 and Karp-002 models claim to be Qwen-3.5 models The Pisces-llm-0206a and Pisces-llm-0206b models claim to be ByteDance models https://t.co/ty55FmNFug Potential new Qwen and ByteDance Seed models are being tested on the Arena The Karp-001 and Karp-002 models claim to be Qwen-3.5 models The Pisces-llm-0206a and" [X Link](https://x.com/AiBattle_/status/2020442493569929720) 2026-02-08T10:20Z [----] followers, 39.3K engagements "GLM-5 has been spotted on Github The next [--] weeks are going to be great" [X Link](https://x.com/AiBattle_/status/2020770939701797098) 2026-02-09T08:05Z [----] followers, 88.3K engagements "MiniMax M2.5 is the first open-weight model to score over 80% on SWE-Bench Verified Minimax-M2.5 SWE-Bench Verified: 80.2% Multi-SWE-Bench: 51.3% BrowseComp: 76.3% https://t.co/oIe9XwZ61R Minimax-M2.5 SWE-Bench Verified: 80.2% Multi-SWE-Bench: 51.3% BrowseComp: 76.3% https://t.co/oIe9XwZ61R" [X Link](https://x.com/AiBattle_/status/2021969766349779238) 2026-02-12T15:28Z [----] followers, 12.8K engagements "The Informations report that we would get next-generation Qwen models and new ByteDance models this month seems to have been correct https://x.com/AiBattle_/status/2016927443051798593s=20 The Information reports new models from ByteDance and Alibaba next month: - ByteDance will launch three new AI models next month - Alibaba will release their next-gen model next month Separately it's likely we're getting the next-gen DeepSeek model next month too Chinese New https://t.co/OBrmqQbc0e https://x.com/AiBattle_/status/2016927443051798593s=20 The Information reports new models from ByteDance and" [X Link](https://x.com/AiBattle_/status/2020444123770028528) 2026-02-08T10:26Z [----] followers, [----] engagements "Rumor: The reason xAI cofounders and team members are leaving is due to pressure from Elon over the lack of progress Another reason is the SpaceX merger which would bring new leadership and additional changes @razroo_chief The rumor is that Elon wasn't happy with the progress they were making and put a lot of pressure on them. Part of it is probably the SpaceX merger; they saw they were going to get new bosses and changes are in the air. I don't know though. I'm watching everyone but very few @razroo_chief The rumor is that Elon wasn't happy with the progress they were making and put a lot of" [X Link](https://x.com/AiBattle_/status/2021461890619277353) 2026-02-11T05:50Z [----] followers, 33.8K engagements "A new DeepSeek model is coming The model appears to be available on the mobile app first. On the mobile app Im seeing a May [----] knowledge cutoff while the website shows a July [----] cutoff date The model is very fast on the app DeepSeek model has updated in app - claims May [----] knowledge cutoff - claims 1M token context window h/t compasslg on dc https://t.co/5HjgpFGmKP DeepSeek model has updated in app - claims May [----] knowledge cutoff - claims 1M token context window h/t compasslg on dc https://t.co/5HjgpFGmKP" [X Link](https://x.com/AiBattle_/status/2021510498714358203) 2026-02-11T09:03Z [----] followers, 41.8K engagements "New DeepSeek update: "DeepSeek Web / APP is currently testing a new long-context model architecture supporting a 1M context window. Note: The API service remains unchanged; it is still V3.2 and only supports a 128K context window. Thank you for your continued support Happy New Year" api https://t.co/Bomzo5EjdU api https://t.co/Bomzo5EjdU" [X Link](https://x.com/AiBattle_/status/2022280288643039235) 2026-02-13T12:02Z [----] followers, 26.5K engagements "Gemini [---] Flash Thinking 24k Prompt: "Create Design a visually striking Tron-style game in a single HTML file where AI-controlled light cycles compete in fast-paced strategic battles against each other"" [X Link](https://x.com/AiBattle_/status/1912954393592205534) 2025-04-17T19:40Z [----] followers, 19.7K engagements "New Gemini model "Claybrook" appeared in Lmarena" [X Link](https://x.com/AiBattle_/status/1913173625613463984) 2025-04-18T10:11Z [----] followers, [----] engagements "Claybrook (Gemini model) doing the rotating square challenge. Claybrook added the ability to reverse the rotation of the square without me asking for it. Never seen that before. Prompt: "Write a p5.js program (in one HTML file) that simulates a few realistically bouncing balls affected by gravity inside a square that rotates around its center. The balls should respond to collisions with the rotating square's walls maintaining physical realism with velocity changes gravity effects and rotation-aware collision detection."" [X Link](https://x.com/AiBattle_/status/1913610879855194360) 2025-04-19T15:09Z [----] followers, [----] engagements "Another new Gemini model "Tomay" dropped in LMarena What is Google cooking 🤔" [X Link](https://x.com/anyuser/status/1913899487455572283) 2025-04-20T10:16Z [----] followers, 30.2K engagements "A new Gemini checkpoint/model "Sunstrike" has appeared in LM Arena" [X Link](https://x.com/anyuser/status/1915848288563302727) 2025-04-25T19:20Z [----] followers, 35.6K engagements "Qwen [--] models were accidentally published on ModelScope but were taken down quickly. We at least now know the Parameters and architectures" [X Link](https://x.com/AiBattle_/status/1916784297819660390) 2025-04-28T09:19Z [----] followers, [----] engagements "Qwen [--] - 235B Benchmark released Its better than o1 o3-mini and R1" [X Link](https://x.com/anyuser/status/1916960357320417350) 2025-04-28T20:58Z [----] followers, [----] engagements "Microsoft is preparing to release a coding model "NextCoder"" [X Link](https://x.com/AiBattle_/status/1918725634152362254) 2025-05-03T17:53Z [----] followers, [----] engagements "There seems to be a new Gemini [---] Pro version on Vertex Ai The current version on Ai studio is from 03-25 while the new one on Vertex Ai is from 05-06" [X Link](https://x.com/anyuser/status/1919735918732222731) 2025-05-06T12:48Z [----] followers, 22.9K engagements "Gemini [---] Pro old and new Benchmark comparison New(left) Old (right)" [X Link](https://x.com/AiBattle_/status/1919788812529439118) 2025-05-06T16:18Z [----] followers, 74.5K engagements "New Gemini [---] Pro Old Gemini [---] Pro - Space Invaders game with one prompt The new Gemini [---] Pro is a solid upgrade for coding tasks. We also already know that Google has an even better model called 'Nightwhisper' likely to be revealed at the Google I/O event" [X Link](https://x.com/AiBattle_/status/1920148282153505267) 2025-05-07T16:06Z [----] followers, [----] engagements "A new Gemini checkpoint/model "Emberwing" has dropped in LM Arena" [X Link](https://x.com/AiBattle_/status/1920417741208453629) 2025-05-08T09:57Z [----] followers, [----] engagements "Gemini [---] Pro Drakesclaw - 3D model of Earth with realistic topography Drakesclaw seems to be somewhere around the [---] Pro tier definitely better than the Emberwing model" [X Link](https://x.com/AiBattle_/status/1921512361795469797) 2025-05-11T10:27Z [----] followers, 35.2K engagements "Since the launch of Gemini [---] Pro on March [--] Google has tested multiple Gemini models / checkpoints on LmArena. The current Gemini [---] Pro is the former "Claybrook" model / checkpoint which first appeared in LmArena on April 18" [X Link](https://x.com/AiBattle_/status/1922373489144545426) 2025-05-13T19:28Z [----] followers, 16.1K engagements "New Google Gemini Model / Checkpoint "Calmriver" in LMarena" [X Link](https://x.com/anyuser/status/1922537564973453753) 2025-05-14T06:20Z [----] followers, [----] engagements "A new Google Gemma model Cutiepie-75 just dropped in LMarena. Google I/O is looking like the biggest AI event of the year so far" [X Link](https://x.com/anyuser/status/1922604744775753953) 2025-05-14T10:47Z [----] followers, 27.8K engagements "The Information recently reported that Anthropic plans to release new Claude Sonnet and Opus models in the coming weeks. Looking at Anthropic's past release patterns they tend to drop a new Claude model every [---] to [--] months. So a Claude [--] release sometime in June seems plausible" [X Link](https://x.com/AiBattle_/status/1923355827089330224) 2025-05-16T12:32Z [----] followers, [----] engagements "All Major Upcoming AI Conferences and Events Next Week" [X Link](https://x.com/anyuser/status/1924143867038752792) 2025-05-18T16:43Z [----] followers, [----] engagements "Gemini [---] Pro Claude Sonnet [--] - 3D mech inspired by Gundam" [X Link](https://x.com/AiBattle_/status/1925600607550767538) 2025-05-22T17:12Z [----] followers, 38.4K engagements "Claude Opus [--] Claude Sonnet [--] - 3D model of the Death Star Both Claude models did really great with this prompt much better than any other model I tried this with" [X Link](https://x.com/AiBattle_/status/1925609042963001795) 2025-05-22T17:45Z [----] followers, 19.6K engagements "2 New Google Gemini models appeared in WebArena "Goldmane" and "Redsword"" [X Link](https://x.com/anyuser/status/1925799869479813417) 2025-05-23T06:24Z [----] followers, 53.1K engagements "Redsword & Goldmane Gemini [---] Pro Nightwhisper and any other Google model for Coding I asked Redsword to create a 3d Mech helmet in one html file it searched and used freely available 3D and HDRI assets and used it to generate the result below. I have never seen a model with such behavior before. Across all the prompts I tested Redsword & Goldmane produced better results compared to [---] Pro. Google cooked with these models" [X Link](https://x.com/anyuser/status/1926189057484128275) 2025-05-24T08:10Z [----] followers, 103.7K engagements "Updated Google Gemini Checkpoint / Model Infographic - May [--] Notable changes: - Confirmation that the Calmriver model is the current Gemini [---] Flash 05-20 -Addition of model Goldmane -Addition of model Redsword" [X Link](https://x.com/anyuser/status/1926305422396321885) 2025-05-24T15:52Z [----] followers, 12K engagements "Interesting behavior from the Gemini model Goldmane When prompting the Goldmane model to generate design concepts or plans it often attempts to include multiple images in its response. This behavior is not present in the Redsword model nor have I seen it in other Gemini models" [X Link](https://x.com/AiBattle_/status/1926323159126487528) 2025-05-24T17:03Z [----] followers, [----] engagements "Redsword Gemini [---] Pro - Space Invaders I have ran this prompt with nearly every Gemini checkpoint / model so far Redsword produced the most complete (gameplay audio) and visually pleasing result i have seen till now" [X Link](https://x.com/anyuser/status/1926566590599831974) 2025-05-25T09:10Z [----] followers, 31.4K engagements "New minor Deepseek R1 update Deepseek has released a minor update to their R1 model now live. Notably the Chain-of-Thought (CoT) behavior appears to have changed significantly" [X Link](https://x.com/anyuser/status/1927695263717601425) 2025-05-28T11:55Z [----] followers, 81.6K engagements "First benchmark for the new Deepseek R1 The new Deepseek R1-0528 performs nearly on par with o3 (High) on the LiveCodeBench benchmark" [X Link](https://x.com/anyuser/status/1927824419478536405) 2025-05-28T20:28Z [----] followers, 16.8K engagements "Kingsfall - Website that showcases a AAA video game" [X Link](https://x.com/anyuser/status/1930509649981141326) 2025-06-05T06:19Z [----] followers, [----] engagements "New Gemini [---] Pro checkpoint is now live in AI Studio" [X Link](https://x.com/AiBattle_/status/1930650346054950959) 2025-06-05T15:38Z [----] followers, [----] engagements "The new Gemini [---] Pro 06-05 dethrones Claude Opus [--] on the WebDev Arena Leaderboard" [X Link](https://x.com/AiBattle_/status/1930656164510871568) 2025-06-05T16:01Z [----] followers, [----] engagements "Kingsfall - 3D simulation of a rocket leaving Earth" [X Link](https://x.com/AiBattle_/status/1930926978544083446) 2025-06-06T09:57Z [----] followers, [----] engagements "o3-Pro (high) performs worse than o3 (high) on ARC-AGI-1 and ARC-AGI-2" [X Link](https://x.com/anyuser/status/1932541694878073272) 2025-06-10T20:53Z [----] followers, 11.4K engagements "New Google Gemini model "Blacktooth" in LMArena" [X Link](https://x.com/anyuser/status/1933883903816700134) 2025-06-14T13:47Z [----] followers, 54.3K engagements "Google's Veo [--] has been dethroned by two models on the Artificial Analysis Image-to-Video Leaderboard. It hasn't even been a month since Veo [--] was released" [X Link](https://x.com/anyuser/status/1934601196054155302) 2025-06-16T13:17Z [----] followers, 47.9K engagements "Kingfall VS Blacktooth - Gen [--] Starter Pokmon SVG" [X Link](https://x.com/anyuser/status/1935339687977353672) 2025-06-18T14:11Z [----] followers, [----] engagements "Another new Google Gemini model "Stonebloom" has dropped in WebArena" [X Link](https://x.com/anyuser/status/1936427533962244170) 2025-06-21T14:14Z [----] followers, 22.4K engagements "Google Gemini Checkpoint / Model Summary June" [X Link](https://x.com/AiBattle_/status/1939986349135626749) 2025-07-01T09:56Z [----] followers, [----] engagements "Mentions of [--] Grok [--] models found in the source code of the xAI console. Grok [--] and Grok [--] Code Grok 4: - Our latest and greatest flagship model offering unparalleled performance in natural language math and reasoning the perfect jack of all trades Grok [--] Code: - A model purpose built to be your coding companion. Ask it questions about your code or embed directly into your code editor" [X Link](https://x.com/anyuser/status/1940139539525419512) 2025-07-01T20:04Z [----] followers, 157.9K engagements "A new Google Gemini model "Wolfstride" is being tested in LmArena" [X Link](https://x.com/AiBattle_/status/1941051274268758243) 2025-07-04T08:27Z [----] followers, 22.6K engagements "When Claude [--] was announced Dario mentioned that minor version updates for the Claude [--] series might be released more frequently. BREAKING 🚨: Some users who have received access to "Claude Neptune v3" are reporting that it can consistently solve math problems at a level of o3 Pro and "Kingfall". The next leap 👀 h/t @No_name_890098 https://t.co/HmMNoxDMns BREAKING 🚨: Some users who have received access to "Claude Neptune v3" are reporting that it can consistently solve math problems at a level of o3 Pro and "Kingfall". The next leap 👀 h/t @No_name_890098 https://t.co/HmMNoxDMns" [X Link](https://x.com/anyuser/status/1941954216781566438) 2025-07-06T20:15Z [----] followers, 45.6K engagements "Grok [--] - 3D Simulation of a Spaceship landing on Mars" [X Link](https://x.com/AiBattle_/status/1943193295611437143) 2025-07-10T06:19Z [----] followers, 45K engagements ""Kingfall" felt like the best and most consistent Google model I have tried so far. This little basketball game was created with a very simple prompt on the first attempt BREAKING 🚨: Google is preparing to release Deep Think on Gemini in the coming weeks and working on a new Agent Mode Deep Think on Gemini performs very close to the leaked "Kingfall" model. What's Agent Mode Check below 👀 https://t.co/kuppTgkjXA BREAKING 🚨: Google is preparing to release Deep Think on Gemini in the coming weeks and working on a new Agent Mode Deep Think on Gemini performs very close to the leaked" [X Link](https://x.com/AiBattle_/status/1943609365178536272) 2025-07-11T09:52Z [----] followers, 12.9K engagements "Kimi K2 - 3D model of an AK-47 K2's habit of adding particle effects reminds me of the Claude models" [X Link](https://x.com/AiBattle_/status/1943695369457541527) 2025-07-11T15:34Z [----] followers, [----] engagements "Kimi K2 non-reasoning is already great at coding and creative writing can't wait to see the performance of K2 with reasoning" [X Link](https://x.com/anyuser/status/1944773830926274822) 2025-07-14T14:59Z [----] followers, 13.8K engagements "Google Gemini Checkpoint / Model Summary July update Changes: - The model "Wolfstride" was added on July 4" [X Link](https://x.com/AiBattle_/status/1945457087250575452) 2025-07-16T12:14Z [----] followers, 11.3K engagements "OpenAI is testing a new model called "o3-alpha-responses-2025-07-17" on WebArena The model will appear with the name "Anonymous-Chatbot"" [X Link](https://x.com/anyuser/status/1946106642598162922) 2025-07-18T07:15Z [----] followers, 253.6K engagements "Space Invaders game from the new o3 model 👇" [X Link](https://x.com/AiBattle_/status/1946117069000392840) 2025-07-18T07:57Z [----] followers, [----] engagements "o3-Alpha Kingfall o3-Alpha is OpenAIs most capable coding model yet on par with the best from Anthropic and Google. The level of detail and functionality it adds to even very simple prompts is impressive. Its hard to believe that o3-Alpha is just a new checkpoint for o3 considering the jump in performance" [X Link](https://x.com/anyuser/status/1946500208344649980) 2025-07-19T09:19Z [----] followers, 39K engagements "We could see GPT-5 released before September OpenAI pulled the "o3-Alpha" model from public testing just [--] hours after it went live maybe an indication that a full launch is close. In the past when OpenAI tested secret models like "Optimus Alpha" and "Quasar Alpha" the official releases followed in [--] days for Quasar and [--] days for Optimus Alpha. Heard GPT-5 is imminent from a little bird. - Its not one model but multiple models. It has a router that switches between reasoning non-reasoning and tool-using models. - Thats why Sam said theyd fix model naming: prompts will just auto-route to" [X Link](https://x.com/anyuser/status/1946862009284206608) 2025-07-20T09:17Z [----] followers, 30.8K engagements "Kimi K2 Qwen-3-235B-A22B-2507 The new updated Qwen [--] model beats Kimi K2 on most benchmarks. The jump on the ARC-AGI score is especially impressive An updated reasoning model is also on the way according to Qwen researchers" [X Link](https://x.com/anyuser/status/1947356300309860559) 2025-07-21T18:01Z [----] followers, 108K engagements "A potential new Google Gemini model "Nightride-on" has entered LmArena The naming scheme this time seems quite different from the usual naming scheme Google uses for its secret models" [X Link](https://x.com/AiBattle_/status/1947783931555709204) 2025-07-22T22:20Z [----] followers, [----] engagements "The newly added 'Lobster' model on WebdevArena is likely part of OpenAI's o3-Alpha series When I used my mini basketball game prompt the output was similar to what I got from o3-Alpha but a worse version of it" [X Link](https://x.com/anyuser/status/1948686031319957784) 2025-07-25T10:05Z [----] followers, 13K engagements "Qwen's newest reasoning model is on the same tier as OpenAI's o3 at least according to benchmarks" [X Link](https://x.com/anyuser/status/1948695733885903185) 2025-07-25T10:44Z [----] followers, 23.5K engagements "2 new potential OpenAI models have entered LmArena The Zenith model in particular seems really good outperforming the o3-Alpha model on one of my test prompts. It also tends to generate lengthy detailed code" [X Link](https://x.com/anyuser/status/1948871083198693501) 2025-07-25T22:20Z [----] followers, 13.7K engagements "Zenith - Doom Game The Zenith model that is being tested in LmArena is producing some amazing outputs With just a single prompt it generated gun sounds sprinting mechanics a minimap and detailed textures for a Doom-style game" [X Link](https://x.com/anyuser/status/1949048047398240563) 2025-07-26T10:04Z [----] followers, 26.6K engagements "Wan [---] has just been released on Hugging Face. It introduces a Mixture-of-Experts (MoE) architecture into video diffusion models On Wan-Bench [---] Wan [---] outperforms other leading models" [X Link](https://x.com/AiBattle_/status/1949810186543132744) 2025-07-28T12:32Z [----] followers, [----] engagements "Horizon Alpha - Webdev prompts The Horizon Alpha model seems decent at Webdev design especially considering it's likely a small model. It definitely has its own unique design style" [X Link](https://x.com/AiBattle_/status/1950881201478140197) 2025-07-31T11:28Z [----] followers, 40.6K engagements "Yesterday an exploit was shared on Telegram that allowed users to access GPT-5 through Perplexity The release of GPT-5 in the coming week seems increasingly likely" [X Link](https://x.com/anyuser/status/1951948640353677578) 2025-08-03T10:09Z [----] followers, 22.5K engagements "Anthropic has released Claude Opus [---] Key Improvements: Coding: - 74.5% on SWE-bench Verified state-of-the-art performance - Stronger multi-file code refactoring and precise debugging (noted by GitHub and Rakuten). Reasoning & Agentic Tasks: - Better real-world coding agentic search and detail tracking. Availability & Pricing: -Live for paid Claude users Claude Code API Amazon Bedrock and Vertex AI -Same pricing as Opus" [X Link](https://x.com/anyuser/status/1952769692793278781) 2025-08-05T16:32Z [----] followers, [----] engagements "OpenAI's open-source models are not Horizon Alpha or Beta I ran the same prompts again and the results from OSS-120 were significantly worse than those produced by the Horizon models Horizon Alpha - Webdev prompts The Horizon Alpha model seems decent at Webdev design especially considering it's likely a small model. It definitely has its own unique design style https://t.co/2zWGSoWOOC Horizon Alpha - Webdev prompts The Horizon Alpha model seems decent at Webdev design especially considering it's likely a small model. It definitely has its own unique design style https://t.co/2zWGSoWOOC" [X Link](https://x.com/AiBattle_/status/1952783964042633426) 2025-08-05T17:29Z [----] followers, 73.9K engagements "Now that we know that Genie [--] supports video input someone with access to it should make this game playable Something we discovered by accident: what happens if we start Genie [--] from a video and a completely unrelated prompt Turns out the model really really wants to make it work to the point where it emulates itself. The prompt in this one is about a trex on a tropical island. https://t.co/XCrmGVGLnR Something we discovered by accident: what happens if we start Genie [--] from a video and a completely unrelated prompt Turns out the model really really wants to make it work to the point where it" [X Link](https://x.com/anyuser/status/1953196617131045353) 2025-08-06T20:48Z [----] followers, 11.7K engagements "GPT-5 (High) scores 9.9% on ARC-AGI-2 Grok [--] (Thinking) scored 16.0%" [X Link](https://x.com/anyuser/status/1953508582927778188) 2025-08-07T17:28Z [----] followers, [----] engagements "GPT-5 - Goblin FPS Game GPT-5 is a clear step up from o3 in coding and multi-turn workflows With [----] requests per week and the option to use your ChatGPT account Codex CLI could compete with Claude Code especially as OpenAI keeps improving GPT-5" [X Link](https://x.com/anyuser/status/1955218761750818854) 2025-08-12T10:44Z [----] followers, 20.4K engagements "2 new models just dropped on LmArena both identify as DeepSeek models : very-secret-and-fun-model and highly-classified-and-cheerful-bot."" [X Link](https://x.com/AiBattle_/status/1957608412168192084) 2025-08-19T00:59Z [----] followers, 20.7K engagements "A new mystery model has appeared in Cursor called "Sonic." It is likely a version of Grok Code or the Grok [----] model Its performance is not that great but it is very fast which suggests it might be a smaller mini version" [X Link](https://x.com/anyuser/status/1957994122876235842) 2025-08-20T02:32Z [----] followers, [----] engagements "4 new models have dropped in LmArena: Yosemite - Claims to be a Grok model (maybe the Grok Code model) Clippy - Claims to be a Microsoft model Millennium - Claims to be a Microsoft model Vista - Claims to be a Microsoft model" [X Link](https://x.com/anyuser/status/1959024882223661450) 2025-08-22T22:48Z [----] followers, 21.1K engagements "Flash-2.5-Flash-Image (Nano-Banana) leads both the Text-to-Image and Image-Edit leaderboards on LmArena holding a big lead over Flux-1-kontext-Max in Image-Edit" [X Link](https://x.com/anyuser/status/1960348087546724623) 2025-08-26T14:26Z [----] followers, [----] engagements "A potential new OpenAI model King-Kedra-0827 is being tested on WebdevArena. After the voting process the OpenAI logo appears beside the name and the model identifies itself as an OpenAI model OpenAI has tested a few models on WebdevArena before with Anonymous-Chatbot-0717 the most similar by naming scheme both names ending in a numeric suffix that likely marks the checkpoint date" [X Link](https://x.com/anyuser/status/1961675430491992391) 2025-08-30T06:20Z [----] followers, 31.6K engagements "A new Image model "DH3" is currently being tested on the Artificial Analysis Image Arena From initial testing the model seems to be on par with Nano-Banana at Image editing According to @bdsqlsz there are two Chinese models one open-weights and one proprietary that are about to be announced and are comparable to Nano-Banana Hopefully this is the open-weights one It's a bit crazy there are two image models about to be announced that are comparable to the nano banana.😋 It's a bit crazy there are two image models about to be announced that are comparable to the nano banana.😋" [X Link](https://x.com/AiBattle_/status/1963871389447758248) 2025-09-05T07:46Z [----] followers, 26.4K engagements "2 new mystery models "Sonoma Sky Alpha" and "Sonoma Dusk Alpha" will soon be dropped on Openrouter" [X Link](https://x.com/anyuser/status/1964063307624542492) 2025-09-05T20:29Z [----] followers, [----] engagements "New Qwen-3-Next-80B-A3B model incoming "Built on this architecture we trained and open-sourced Qwen3-Next-80B-A3B 80B total parameters only 3B active achieving extreme sparsity and efficiency. Despite its ultra-efficiency it outperforms Qwen3-32B on downstream tasks while requiring **less than 1/10 of the training cost**. Moreover it delivers over **10x higher inference throughput** than Qwen3-32B when handling contexts longer than 32K tokens" https://t.co/HbDdaZOO5L https://t.co/MtS1N60nqR https://t.co/HbDdaZOO5L https://t.co/MtS1N60nqR" [X Link](https://x.com/anyuser/status/1965424992121729185) 2025-09-09T14:40Z [----] followers, 85.2K engagements "A new Google Gemini / Gemma model "Oceanstone" is being tested in LmArena" [X Link](https://x.com/anyuser/status/1967482241753518479) 2025-09-15T06:54Z [----] followers, 177.4K engagements "Another new Google Gemini model "Oceanreef" is being tested in LmArena The model is likely related to the "Oceanstone" model which appeared [--] days ago" [X Link](https://x.com/anyuser/status/1968190145913557049) 2025-09-17T05:47Z [----] followers, 13.7K engagements "Potential new Grok code model "Code-Supernova" is currently being tested in Cursor and Cline" [X Link](https://x.com/anyuser/status/1969319488093892925) 2025-09-20T08:35Z [----] followers, 16.6K engagements "New Qwen3-Omni-7B model is soon to be released" [X Link](https://x.com/anyuser/status/1969467034976096294) 2025-09-20T18:21Z [----] followers, 49.6K engagements "New Deepseek v3 version "Terminus" incoming" [X Link](https://x.com/AiBattle_/status/1970092946411425997) 2025-09-22T11:48Z [----] followers, 56.7K engagements "Potential new Grok model "Rainbow" is currently being tested in WebDevArena" [X Link](https://x.com/AiBattle_/status/1971138181174292925) 2025-09-25T09:02Z [----] followers, [----] engagements "New Checkpoints for Gemini [---] Flash and Flash-Lite This latest [---] Flash model comes with improvements in two key areas we heard consistent feedback on: - Better agentic tool use: We've improved how the model uses tools leading to better performance in more complex agentic and multi-step applications This model shows noticeable improvements on key agentic benchmarks including a 5% gain on SWE-Bench Verified compared to our last release (48.9% 54%) - More efficient: With thinking on the model is now significantly more cost-efficientachieving higher quality outputs while using fewer tokens" [X Link](https://x.com/anyuser/status/1971265537184252376) 2025-09-25T17:28Z [----] followers, 14.4K engagements "New Image model "Lavender" is currently being tested on the Artificial Analysis Arena The model is likely related to the GPT-image-1 model possibly an update Prompt: A cartoonish 1940s detectives office rendered in a loose watercolor style: a gumshoe in a pinstriped suit fedora tilted low studies a case file beneath a single incandescent bulb. Smoke curls from his cigarette and old newspapers a rotary phone and a half-empty whiskey glass set the moody stage" [X Link](https://x.com/anyuser/status/1974129350548050232) 2025-10-03T15:08Z [----] followers, 21.6K engagements "Gemini [--] Pro () Zenith (GPT-5-Checkpoint) - Doom Game This is probably the best result so far from a Gemini model for this prompt However it still lacks some of the functionality that Zenith had such as the minimap sound effects and enhanced health UI The enemies though look significantly better in what the Gemini model produced" [X Link](https://x.com/anyuser/status/1974827402854281619) 2025-10-05T13:21Z [----] followers, 22K engagements "Gemini [--] "ecpt" Gemini [---] Pro - Space Invaders I used to get significantly better results from Gemini [---] Pro but its performance seems to have declined on this prompt The output shown in the video is the best I got after multiple tries while Gemini 3s result is from the first try" [X Link](https://x.com/anyuser/status/1977338897483841621) 2025-10-12T11:41Z [----] followers, 96.8K engagements "Gemini [--] Pro related strings found on the Gemini site. The release should be really close now the strings dont lie "We've upgraded you from the previous model to [---] Pro our smartest model yet." the strings dont lie "We've upgraded you from the previous model to [---] Pro our smartest model yet."" [X Link](https://x.com/AiBattle_/status/1978540060447293611) 2025-10-15T19:14Z [----] followers, 20.2K engagements "2 new Google Gemini models "Orionmist" & "Lithiumflow" are currently being tested in LmArena" [X Link](https://x.com/anyuser/status/1979975338177282397) 2025-10-19T18:18Z [----] followers, 54.6K engagements "Gemini [--] "9d30" Lithiumflow Orionmist Gemini [---] Pro - 3D Voxel first gen starter Pokemon scene Lithiumflow and Orionmist seem slightly less consistent and overall weaker than some of the Gemini [--] checkpoints Ive tested on AI Studio Still they demonstrate a significant improvement over Gemini [---] Pro" [X Link](https://x.com/anyuser/status/1981665288299966689) 2025-10-24T10:13Z [----] followers, 19.7K engagements "Oak Willow - Xbox Controller SVG These [--] mystery models are likely different GPT-5 checkpoints 🚨 Breaking News DesignArena [--] New Mystery Models willow cedar birch oak You can get this model superfast compared to Lmarena in designarena link in comment and also these models are from big lab of USA iykyk 😉 https://t.co/ha7e2Uedgs 🚨 Breaking News DesignArena [--] New Mystery Models willow cedar birch oak You can get this model superfast compared to Lmarena in designarena link in comment and also these models are from big lab of USA iykyk 😉 https://t.co/ha7e2Uedgs" [X Link](https://x.com/AiBattle_/status/1984580272248246750) 2025-11-01T11:16Z [----] followers, 28.8K engagements "New GPT-5 checkpoints have been added to DesignArena The reasoning budget of the new models: Firefly: [--] Chrysalis: [--] Cicada: [--] Caterpillar: [---] The new checkpoints seem a bit better than the old ones from initial testing 🚨 Breaking News - DesignArena [--] New Mystery Models cicada caterpillar chrysalis firefly You can get this model superfast compared to Lmarena in designarena link in comment and also these models are from big lab of USA😉 https://t.co/dCh6DFgyTf 🚨 Breaking News - DesignArena [--] New Mystery Models cicada caterpillar chrysalis firefly You can get this model superfast compared" [X Link](https://x.com/anyuser/status/1985647870192623831) 2025-11-04T09:58Z [----] followers, 20.1K engagements "The GPT-5 models that were being tested on DesignArena are likely GPT-5-1-Thinking with different reasoning budgets There are also rumors that GPT-5-1-Thinking will have a bigger context window and reduced API pricing LEAK: GPT-5-1 Thinking officially confirmed by OpenAI https://t.co/lLzvZcMVmT LEAK: GPT-5-1 Thinking officially confirmed by OpenAI https://t.co/lLzvZcMVmT" [X Link](https://x.com/anyuser/status/1986375845494071686) 2025-11-06T10:11Z [----] followers, 15.2K engagements "Kimi-K2-Thinking beats GPT-5-Pro on Humanitys Last exam Kimi-K2-Thinking-Heavy achieves over 50% on Humanitys Last exam" [X Link](https://x.com/anyuser/status/1986464995425620207) 2025-11-06T16:05Z [----] followers, 72.3K engagements "Grok Image "Mandarin"Nano-Banana - Last paragraph from "Faust" The new Grok Image model "Mandarin" seems really good text generation capabilities seem to be on par with Nano-Banana Impressive how quickly xAI is improving and iterating on their image and video generation models 🚨 New Grok Image Model on LM Arena: Mandarin Looks pretty solid so far Ill be testing it more soon. Its from xAI. We got some competition for NB2 https://t.co/9Cs1KZlzdj 🚨 New Grok Image Model on LM Arena: Mandarin Looks pretty solid so far Ill be testing it more soon. Its from xAI. We got some competition for NB2" [X Link](https://x.com/anyuser/status/1988187418671886841) 2025-11-11T10:09Z [----] followers, 32.4K engagements "A new Google Gemini model "Riftrunner" is currently being tested on LmArena" [X Link](https://x.com/anyuser/status/1988522687136669995) 2025-11-12T08:22Z [----] followers, 34.3K engagements "Gemini [--] appears to be rolling out now The Canvas feature in the mobile app seems to use the new Gemini [--] model The difference between Web and mobile for the 3D Pokemon voxel scene is huge It seems Gemini [--] is secretely rolling-out on the Gemini app using the Canvas feature Big difference in output quality between web (PC) and mobile versions Left: Web version Right: Mobile version https://t.co/bXPosvA65V It seems Gemini [--] is secretely rolling-out on the Gemini app using the Canvas feature Big difference in output quality between web (PC) and mobile versions Left: Web version Right: Mobile" [X Link](https://x.com/anyuser/status/1988881718627971222) 2025-11-13T08:08Z [----] followers, 118.1K engagements Limited data mode. Full metrics available with subscription: lunarcrush.com/pricing
@AiBattle_ AiBattle
AiBattle posts on X about $googl, prompt, in the, the new the most. They currently have [-----] followers and [---] posts still getting attention that total [------] engagements in the last [--] hours.
Engagements: [------] #
- [--] Week [-------] +47%
- [--] Month [-------] -0.30%
- [--] Months [---------] +201%
Mentions: [--] #
Followers: [-----] #
- [--] Week [-----] +3.80%
- [--] Month [-----] +26%
- [--] Months [-----] +92%
CreatorRank: [-------] #
Social Influence
Social category influence technology brands 35.9% stocks 18.8% gaming 2.56% social networks 1.71% finance 1.71% events 1.71% countries 1.71% celebrities 0.85% products 0.85%
Social topic influence $googl 17.09%, prompt 10.26%, in the 7.69%, the new 7.69%, ai 6.84%, o3 5.98%, open ai 5.13%, bytedance #66, context window #1, anthropic 4.27%
Top accounts mentioned or mentioned by @hibara_ai_lover @aquiffoo @chetaslua @v0 @scaling01 @razroochief @noname890098 @bdsqlsz @tonitrades_ @aero96193997 @aoligei15 @slartibart_ @girish_lelouch @drbeavisai @gauravdhiman_ai @apathium65906 @xiaotiaowang @streyai @wokoma_festus @satoshissss
Top assets mentioned Alphabet Inc Class A (GOOGL) Microsoft Corp. (MSFT) Trex Company, Inc. (TREX)
Top Social Posts
Top posts by engagements in the last [--] hours
"New Grok image model is being tested on LmArena under the name "Sumo""
X Link 2025-12-30T11:54Z [----] followers, 25.3K engagements
"ByteDance has been testing its new Doubao model for a week now in Kilo Code under the name "Giga-Potato" Description from Kilo Code: "In our internal benchmarks it has outperformed nearly every open-weight model weve tested on long-context coding tasks - Context Window: 256k Tokens - Max Output: 32k Tokens - Strict Adherence: The model is showing exceptional discipline in following system prompts making it ideal for enterprise environments with strict linting and style guidelines" https://twitter.com/i/web/status/2014361796279181388 https://twitter.com/i/web/status/2014361796279181388"
X Link 2026-01-22T15:37Z [----] followers, 15.4K engagements
"Kimi K2.5 is live on the Kimi website The model looks really promising on zero-shot coding prompts so far. Well see how it translates to agentic coding tasks but so far Im really excited"
X Link 2026-01-26T21:39Z [----] followers, 50.2K engagements
"Kimi K2.5 scores 46.8% on SimpleBench DeepSeek V3.2 Speciale remains the open-weights model with the highest score on this benchmark at 52.6%"
X Link 2026-01-28T13:30Z [----] followers, 15.8K engagements
"New Claude model update(s) are coming The upcoming "Fennec" model (Sonnet update) seems to be better than Opus [---] according to tests from @chetaslua Big week for Anthropic fans coming up😉 (Or perhaps just anyone who uses AI to code) Big week for Anthropic fans coming up😉 (Or perhaps just anyone who uses AI to code)"
X Link 2026-01-31T15:24Z [----] followers, 224.7K engagements
"Newly released Stepfun model "Step-3.5-Flash" beats DeepSeek v3.2 on several benchmarks while having far fewer parameters Step-3.5-Flash: 196B total / 11B active Parameters DeepSeek v3.2: 671B total / 37B active Parameters This week / month will likely have some of the most impactful model releases in a while from both U.S. and Chinese labs Exciting days ahead https://twitter.com/i/web/status/2018143041840697520 https://twitter.com/i/web/status/2018143041840697520"
X Link 2026-02-02T02:02Z [----] followers, 40.8K engagements
"@v0 just deleted the Sonnet [--] tweet @scaling01 was right it was just engagement baiting"
X Link 2026-02-04T14:51Z [----] followers, [----] engagements
"Opus [---] has been found in the Perplexity API "Anthropic's most advanced model" 🚨NEW: Claude Opus [---] & Claude Opus [---] Thinking are now live on Perplexity's APIs Looks like we're getting it today and Sonnet [--] later https://t.co/pL3m2yOyUd https://t.co/glaky9eAAP 🚨NEW: Claude Opus [---] & Claude Opus [---] Thinking are now live on Perplexity's APIs Looks like we're getting it today and Sonnet [--] later https://t.co/pL3m2yOyUd https://t.co/glaky9eAAP"
X Link 2026-02-05T12:19Z [----] followers, 29.5K engagements
"Claude Opus [---] has the same pricing as Claude Opus 4.5"
X Link 2026-02-05T17:43Z [----] followers, 23.3K engagements
"Benchmark scores for GPT-5.3-Codex (xhigh)"
X Link 2026-02-05T18:08Z [----] followers, [----] engagements
"Claude Opus [---] scores 67.6% on Simple-Bench"
X Link 2026-02-06T08:37Z [----] followers, [----] engagements
"The Information reports new models from ByteDance and Alibaba next month: - ByteDance will launch three new AI models next month - Alibaba will release their next-gen model next month Separately it's likely we're getting the next-gen DeepSeek model next month too Chinese New Year is going to be crazy ByteDance and Alibaba are set to launch new AI models (The Information). ByteDance plans to unveil three new AI models next month and Alibaba is also expected to introduce its next-generation AI model next month. ByteDance and Alibaba are set to launch new AI models (The Information). ByteDance"
X Link 2026-01-29T17:32Z [----] followers, 28.3K engagements
"Potential new Qwen and ByteDance Seed models are being tested on the Arena The Karp-001 and Karp-002 models claim to be Qwen-3.5 models The Pisces-llm-0206a and Pisces-llm-0206b models claim to be ByteDance models"
X Link 2026-02-07T11:46Z [----] followers, 45.8K engagements
"GLM-5 scores higher than Gemini [--] Pro on the Artificial Analysis Intelligence Index GLM-5 is now the open-weight model with the highest score"
X Link 2026-02-11T19:08Z [----] followers, 45.5K engagements
"Opus [---] performs worse than Opus [---] on SWE-Bench"
X Link 2026-02-05T17:47Z [----] followers, 84.8K engagements
"Claude Opus [---] ARC-AGI scores are live Opus [---] (120k High) has the highest score: ARC-AGI-1: 94.00% ARC-AGI-2: 69.20%"
X Link 2026-02-05T18:52Z [----] followers, [----] engagements
"It claims to be a Claude model. Anthropic doesnt do stealth model releases as far as I know so its likely from a Chinese lab It has a 200k context window the same as the GLM [---] and [---] models so it might be a new GLM model The GLM models are also known for sometimes identifying themselves as Claude models https://twitter.com/i/web/status/2019836771484184862 https://twitter.com/i/web/status/2019836771484184862"
X Link 2026-02-06T18:13Z [----] followers, [----] engagements
"A new stealth model "Pony-Alpha" is being tested on OpenRouter"
X Link 2026-02-06T17:50Z [----] followers, 13.1K engagements
"Pelican SVG looking good"
X Link 2026-02-06T17:53Z [----] followers, [----] engagements
"Qwen [---] has been spotted on GitHub - Qwen3.5-9B-Instruct - Qwen3.5-35B-A3B-Instruct The [--] currently available models on the Arena "Karp-001" and "Karp-002" could possibly be the small Qwen-3.5 models Potential new Qwen and ByteDance Seed models are being tested on the Arena The Karp-001 and Karp-002 models claim to be Qwen-3.5 models The Pisces-llm-0206a and Pisces-llm-0206b models claim to be ByteDance models https://t.co/ty55FmNFug Potential new Qwen and ByteDance Seed models are being tested on the Arena The Karp-001 and Karp-002 models claim to be Qwen-3.5 models The Pisces-llm-0206a and"
X Link 2026-02-08T10:20Z [----] followers, 39.3K engagements
"GLM-5 has been spotted on Github The next [--] weeks are going to be great"
X Link 2026-02-09T08:05Z [----] followers, 88.3K engagements
"MiniMax M2.5 is the first open-weight model to score over 80% on SWE-Bench Verified Minimax-M2.5 SWE-Bench Verified: 80.2% Multi-SWE-Bench: 51.3% BrowseComp: 76.3% https://t.co/oIe9XwZ61R Minimax-M2.5 SWE-Bench Verified: 80.2% Multi-SWE-Bench: 51.3% BrowseComp: 76.3% https://t.co/oIe9XwZ61R"
X Link 2026-02-12T15:28Z [----] followers, 12.8K engagements
"The Informations report that we would get next-generation Qwen models and new ByteDance models this month seems to have been correct https://x.com/AiBattle_/status/2016927443051798593s=20 The Information reports new models from ByteDance and Alibaba next month: - ByteDance will launch three new AI models next month - Alibaba will release their next-gen model next month Separately it's likely we're getting the next-gen DeepSeek model next month too Chinese New https://t.co/OBrmqQbc0e https://x.com/AiBattle_/status/2016927443051798593s=20 The Information reports new models from ByteDance and"
X Link 2026-02-08T10:26Z [----] followers, [----] engagements
"Rumor: The reason xAI cofounders and team members are leaving is due to pressure from Elon over the lack of progress Another reason is the SpaceX merger which would bring new leadership and additional changes @razroo_chief The rumor is that Elon wasn't happy with the progress they were making and put a lot of pressure on them. Part of it is probably the SpaceX merger; they saw they were going to get new bosses and changes are in the air. I don't know though. I'm watching everyone but very few @razroo_chief The rumor is that Elon wasn't happy with the progress they were making and put a lot of"
X Link 2026-02-11T05:50Z [----] followers, 33.8K engagements
"A new DeepSeek model is coming The model appears to be available on the mobile app first. On the mobile app Im seeing a May [----] knowledge cutoff while the website shows a July [----] cutoff date The model is very fast on the app DeepSeek model has updated in app - claims May [----] knowledge cutoff - claims 1M token context window h/t compasslg on dc https://t.co/5HjgpFGmKP DeepSeek model has updated in app - claims May [----] knowledge cutoff - claims 1M token context window h/t compasslg on dc https://t.co/5HjgpFGmKP"
X Link 2026-02-11T09:03Z [----] followers, 41.8K engagements
"New DeepSeek update: "DeepSeek Web / APP is currently testing a new long-context model architecture supporting a 1M context window. Note: The API service remains unchanged; it is still V3.2 and only supports a 128K context window. Thank you for your continued support Happy New Year" api https://t.co/Bomzo5EjdU api https://t.co/Bomzo5EjdU"
X Link 2026-02-13T12:02Z [----] followers, 26.5K engagements
"Gemini [---] Flash Thinking 24k Prompt: "Create Design a visually striking Tron-style game in a single HTML file where AI-controlled light cycles compete in fast-paced strategic battles against each other""
X Link 2025-04-17T19:40Z [----] followers, 19.7K engagements
"New Gemini model "Claybrook" appeared in Lmarena"
X Link 2025-04-18T10:11Z [----] followers, [----] engagements
"Claybrook (Gemini model) doing the rotating square challenge. Claybrook added the ability to reverse the rotation of the square without me asking for it. Never seen that before. Prompt: "Write a p5.js program (in one HTML file) that simulates a few realistically bouncing balls affected by gravity inside a square that rotates around its center. The balls should respond to collisions with the rotating square's walls maintaining physical realism with velocity changes gravity effects and rotation-aware collision detection.""
X Link 2025-04-19T15:09Z [----] followers, [----] engagements
"Another new Gemini model "Tomay" dropped in LMarena What is Google cooking 🤔"
X Link 2025-04-20T10:16Z [----] followers, 30.2K engagements
"A new Gemini checkpoint/model "Sunstrike" has appeared in LM Arena"
X Link 2025-04-25T19:20Z [----] followers, 35.6K engagements
"Qwen [--] models were accidentally published on ModelScope but were taken down quickly. We at least now know the Parameters and architectures"
X Link 2025-04-28T09:19Z [----] followers, [----] engagements
"Qwen [--] - 235B Benchmark released Its better than o1 o3-mini and R1"
X Link 2025-04-28T20:58Z [----] followers, [----] engagements
"Microsoft is preparing to release a coding model "NextCoder""
X Link 2025-05-03T17:53Z [----] followers, [----] engagements
"There seems to be a new Gemini [---] Pro version on Vertex Ai The current version on Ai studio is from 03-25 while the new one on Vertex Ai is from 05-06"
X Link 2025-05-06T12:48Z [----] followers, 22.9K engagements
"Gemini [---] Pro old and new Benchmark comparison New(left) Old (right)"
X Link 2025-05-06T16:18Z [----] followers, 74.5K engagements
"New Gemini [---] Pro Old Gemini [---] Pro - Space Invaders game with one prompt The new Gemini [---] Pro is a solid upgrade for coding tasks. We also already know that Google has an even better model called 'Nightwhisper' likely to be revealed at the Google I/O event"
X Link 2025-05-07T16:06Z [----] followers, [----] engagements
"A new Gemini checkpoint/model "Emberwing" has dropped in LM Arena"
X Link 2025-05-08T09:57Z [----] followers, [----] engagements
"Gemini [---] Pro Drakesclaw - 3D model of Earth with realistic topography Drakesclaw seems to be somewhere around the [---] Pro tier definitely better than the Emberwing model"
X Link 2025-05-11T10:27Z [----] followers, 35.2K engagements
"Since the launch of Gemini [---] Pro on March [--] Google has tested multiple Gemini models / checkpoints on LmArena. The current Gemini [---] Pro is the former "Claybrook" model / checkpoint which first appeared in LmArena on April 18"
X Link 2025-05-13T19:28Z [----] followers, 16.1K engagements
"New Google Gemini Model / Checkpoint "Calmriver" in LMarena"
X Link 2025-05-14T06:20Z [----] followers, [----] engagements
"A new Google Gemma model Cutiepie-75 just dropped in LMarena. Google I/O is looking like the biggest AI event of the year so far"
X Link 2025-05-14T10:47Z [----] followers, 27.8K engagements
"The Information recently reported that Anthropic plans to release new Claude Sonnet and Opus models in the coming weeks. Looking at Anthropic's past release patterns they tend to drop a new Claude model every [---] to [--] months. So a Claude [--] release sometime in June seems plausible"
X Link 2025-05-16T12:32Z [----] followers, [----] engagements
"All Major Upcoming AI Conferences and Events Next Week"
X Link 2025-05-18T16:43Z [----] followers, [----] engagements
"Gemini [---] Pro Claude Sonnet [--] - 3D mech inspired by Gundam"
X Link 2025-05-22T17:12Z [----] followers, 38.4K engagements
"Claude Opus [--] Claude Sonnet [--] - 3D model of the Death Star Both Claude models did really great with this prompt much better than any other model I tried this with"
X Link 2025-05-22T17:45Z [----] followers, 19.6K engagements
"2 New Google Gemini models appeared in WebArena "Goldmane" and "Redsword""
X Link 2025-05-23T06:24Z [----] followers, 53.1K engagements
"Redsword & Goldmane Gemini [---] Pro Nightwhisper and any other Google model for Coding I asked Redsword to create a 3d Mech helmet in one html file it searched and used freely available 3D and HDRI assets and used it to generate the result below. I have never seen a model with such behavior before. Across all the prompts I tested Redsword & Goldmane produced better results compared to [---] Pro. Google cooked with these models"
X Link 2025-05-24T08:10Z [----] followers, 103.7K engagements
"Updated Google Gemini Checkpoint / Model Infographic - May [--] Notable changes: - Confirmation that the Calmriver model is the current Gemini [---] Flash 05-20 -Addition of model Goldmane -Addition of model Redsword"
X Link 2025-05-24T15:52Z [----] followers, 12K engagements
"Interesting behavior from the Gemini model Goldmane When prompting the Goldmane model to generate design concepts or plans it often attempts to include multiple images in its response. This behavior is not present in the Redsword model nor have I seen it in other Gemini models"
X Link 2025-05-24T17:03Z [----] followers, [----] engagements
"Redsword Gemini [---] Pro - Space Invaders I have ran this prompt with nearly every Gemini checkpoint / model so far Redsword produced the most complete (gameplay audio) and visually pleasing result i have seen till now"
X Link 2025-05-25T09:10Z [----] followers, 31.4K engagements
"New minor Deepseek R1 update Deepseek has released a minor update to their R1 model now live. Notably the Chain-of-Thought (CoT) behavior appears to have changed significantly"
X Link 2025-05-28T11:55Z [----] followers, 81.6K engagements
"First benchmark for the new Deepseek R1 The new Deepseek R1-0528 performs nearly on par with o3 (High) on the LiveCodeBench benchmark"
X Link 2025-05-28T20:28Z [----] followers, 16.8K engagements
"Kingsfall - Website that showcases a AAA video game"
X Link 2025-06-05T06:19Z [----] followers, [----] engagements
"New Gemini [---] Pro checkpoint is now live in AI Studio"
X Link 2025-06-05T15:38Z [----] followers, [----] engagements
"The new Gemini [---] Pro 06-05 dethrones Claude Opus [--] on the WebDev Arena Leaderboard"
X Link 2025-06-05T16:01Z [----] followers, [----] engagements
"Kingsfall - 3D simulation of a rocket leaving Earth"
X Link 2025-06-06T09:57Z [----] followers, [----] engagements
"o3-Pro (high) performs worse than o3 (high) on ARC-AGI-1 and ARC-AGI-2"
X Link 2025-06-10T20:53Z [----] followers, 11.4K engagements
"New Google Gemini model "Blacktooth" in LMArena"
X Link 2025-06-14T13:47Z [----] followers, 54.3K engagements
"Google's Veo [--] has been dethroned by two models on the Artificial Analysis Image-to-Video Leaderboard. It hasn't even been a month since Veo [--] was released"
X Link 2025-06-16T13:17Z [----] followers, 47.9K engagements
"Kingfall VS Blacktooth - Gen [--] Starter Pokmon SVG"
X Link 2025-06-18T14:11Z [----] followers, [----] engagements
"Another new Google Gemini model "Stonebloom" has dropped in WebArena"
X Link 2025-06-21T14:14Z [----] followers, 22.4K engagements
"Google Gemini Checkpoint / Model Summary June"
X Link 2025-07-01T09:56Z [----] followers, [----] engagements
"Mentions of [--] Grok [--] models found in the source code of the xAI console. Grok [--] and Grok [--] Code Grok 4: - Our latest and greatest flagship model offering unparalleled performance in natural language math and reasoning the perfect jack of all trades Grok [--] Code: - A model purpose built to be your coding companion. Ask it questions about your code or embed directly into your code editor"
X Link 2025-07-01T20:04Z [----] followers, 157.9K engagements
"A new Google Gemini model "Wolfstride" is being tested in LmArena"
X Link 2025-07-04T08:27Z [----] followers, 22.6K engagements
"When Claude [--] was announced Dario mentioned that minor version updates for the Claude [--] series might be released more frequently. BREAKING 🚨: Some users who have received access to "Claude Neptune v3" are reporting that it can consistently solve math problems at a level of o3 Pro and "Kingfall". The next leap 👀 h/t @No_name_890098 https://t.co/HmMNoxDMns BREAKING 🚨: Some users who have received access to "Claude Neptune v3" are reporting that it can consistently solve math problems at a level of o3 Pro and "Kingfall". The next leap 👀 h/t @No_name_890098 https://t.co/HmMNoxDMns"
X Link 2025-07-06T20:15Z [----] followers, 45.6K engagements
"Grok [--] - 3D Simulation of a Spaceship landing on Mars"
X Link 2025-07-10T06:19Z [----] followers, 45K engagements
""Kingfall" felt like the best and most consistent Google model I have tried so far. This little basketball game was created with a very simple prompt on the first attempt BREAKING 🚨: Google is preparing to release Deep Think on Gemini in the coming weeks and working on a new Agent Mode Deep Think on Gemini performs very close to the leaked "Kingfall" model. What's Agent Mode Check below 👀 https://t.co/kuppTgkjXA BREAKING 🚨: Google is preparing to release Deep Think on Gemini in the coming weeks and working on a new Agent Mode Deep Think on Gemini performs very close to the leaked"
X Link 2025-07-11T09:52Z [----] followers, 12.9K engagements
"Kimi K2 - 3D model of an AK-47 K2's habit of adding particle effects reminds me of the Claude models"
X Link 2025-07-11T15:34Z [----] followers, [----] engagements
"Kimi K2 non-reasoning is already great at coding and creative writing can't wait to see the performance of K2 with reasoning"
X Link 2025-07-14T14:59Z [----] followers, 13.8K engagements
"Google Gemini Checkpoint / Model Summary July update Changes: - The model "Wolfstride" was added on July 4"
X Link 2025-07-16T12:14Z [----] followers, 11.3K engagements
"OpenAI is testing a new model called "o3-alpha-responses-2025-07-17" on WebArena The model will appear with the name "Anonymous-Chatbot""
X Link 2025-07-18T07:15Z [----] followers, 253.6K engagements
"Space Invaders game from the new o3 model 👇"
X Link 2025-07-18T07:57Z [----] followers, [----] engagements
"o3-Alpha Kingfall o3-Alpha is OpenAIs most capable coding model yet on par with the best from Anthropic and Google. The level of detail and functionality it adds to even very simple prompts is impressive. Its hard to believe that o3-Alpha is just a new checkpoint for o3 considering the jump in performance"
X Link 2025-07-19T09:19Z [----] followers, 39K engagements
"We could see GPT-5 released before September OpenAI pulled the "o3-Alpha" model from public testing just [--] hours after it went live maybe an indication that a full launch is close. In the past when OpenAI tested secret models like "Optimus Alpha" and "Quasar Alpha" the official releases followed in [--] days for Quasar and [--] days for Optimus Alpha. Heard GPT-5 is imminent from a little bird. - Its not one model but multiple models. It has a router that switches between reasoning non-reasoning and tool-using models. - Thats why Sam said theyd fix model naming: prompts will just auto-route to"
X Link 2025-07-20T09:17Z [----] followers, 30.8K engagements
"Kimi K2 Qwen-3-235B-A22B-2507 The new updated Qwen [--] model beats Kimi K2 on most benchmarks. The jump on the ARC-AGI score is especially impressive An updated reasoning model is also on the way according to Qwen researchers"
X Link 2025-07-21T18:01Z [----] followers, 108K engagements
"A potential new Google Gemini model "Nightride-on" has entered LmArena The naming scheme this time seems quite different from the usual naming scheme Google uses for its secret models"
X Link 2025-07-22T22:20Z [----] followers, [----] engagements
"The newly added 'Lobster' model on WebdevArena is likely part of OpenAI's o3-Alpha series When I used my mini basketball game prompt the output was similar to what I got from o3-Alpha but a worse version of it"
X Link 2025-07-25T10:05Z [----] followers, 13K engagements
"Qwen's newest reasoning model is on the same tier as OpenAI's o3 at least according to benchmarks"
X Link 2025-07-25T10:44Z [----] followers, 23.5K engagements
"2 new potential OpenAI models have entered LmArena The Zenith model in particular seems really good outperforming the o3-Alpha model on one of my test prompts. It also tends to generate lengthy detailed code"
X Link 2025-07-25T22:20Z [----] followers, 13.7K engagements
"Zenith - Doom Game The Zenith model that is being tested in LmArena is producing some amazing outputs With just a single prompt it generated gun sounds sprinting mechanics a minimap and detailed textures for a Doom-style game"
X Link 2025-07-26T10:04Z [----] followers, 26.6K engagements
"Wan [---] has just been released on Hugging Face. It introduces a Mixture-of-Experts (MoE) architecture into video diffusion models On Wan-Bench [---] Wan [---] outperforms other leading models"
X Link 2025-07-28T12:32Z [----] followers, [----] engagements
"Horizon Alpha - Webdev prompts The Horizon Alpha model seems decent at Webdev design especially considering it's likely a small model. It definitely has its own unique design style"
X Link 2025-07-31T11:28Z [----] followers, 40.6K engagements
"Yesterday an exploit was shared on Telegram that allowed users to access GPT-5 through Perplexity The release of GPT-5 in the coming week seems increasingly likely"
X Link 2025-08-03T10:09Z [----] followers, 22.5K engagements
"Anthropic has released Claude Opus [---] Key Improvements: Coding: - 74.5% on SWE-bench Verified state-of-the-art performance - Stronger multi-file code refactoring and precise debugging (noted by GitHub and Rakuten). Reasoning & Agentic Tasks: - Better real-world coding agentic search and detail tracking. Availability & Pricing: -Live for paid Claude users Claude Code API Amazon Bedrock and Vertex AI -Same pricing as Opus"
X Link 2025-08-05T16:32Z [----] followers, [----] engagements
"OpenAI's open-source models are not Horizon Alpha or Beta I ran the same prompts again and the results from OSS-120 were significantly worse than those produced by the Horizon models Horizon Alpha - Webdev prompts The Horizon Alpha model seems decent at Webdev design especially considering it's likely a small model. It definitely has its own unique design style https://t.co/2zWGSoWOOC Horizon Alpha - Webdev prompts The Horizon Alpha model seems decent at Webdev design especially considering it's likely a small model. It definitely has its own unique design style https://t.co/2zWGSoWOOC"
X Link 2025-08-05T17:29Z [----] followers, 73.9K engagements
"Now that we know that Genie [--] supports video input someone with access to it should make this game playable Something we discovered by accident: what happens if we start Genie [--] from a video and a completely unrelated prompt Turns out the model really really wants to make it work to the point where it emulates itself. The prompt in this one is about a trex on a tropical island. https://t.co/XCrmGVGLnR Something we discovered by accident: what happens if we start Genie [--] from a video and a completely unrelated prompt Turns out the model really really wants to make it work to the point where it"
X Link 2025-08-06T20:48Z [----] followers, 11.7K engagements
"GPT-5 (High) scores 9.9% on ARC-AGI-2 Grok [--] (Thinking) scored 16.0%"
X Link 2025-08-07T17:28Z [----] followers, [----] engagements
"GPT-5 - Goblin FPS Game GPT-5 is a clear step up from o3 in coding and multi-turn workflows With [----] requests per week and the option to use your ChatGPT account Codex CLI could compete with Claude Code especially as OpenAI keeps improving GPT-5"
X Link 2025-08-12T10:44Z [----] followers, 20.4K engagements
"2 new models just dropped on LmArena both identify as DeepSeek models : very-secret-and-fun-model and highly-classified-and-cheerful-bot.""
X Link 2025-08-19T00:59Z [----] followers, 20.7K engagements
"A new mystery model has appeared in Cursor called "Sonic." It is likely a version of Grok Code or the Grok [----] model Its performance is not that great but it is very fast which suggests it might be a smaller mini version"
X Link 2025-08-20T02:32Z [----] followers, [----] engagements
"4 new models have dropped in LmArena: Yosemite - Claims to be a Grok model (maybe the Grok Code model) Clippy - Claims to be a Microsoft model Millennium - Claims to be a Microsoft model Vista - Claims to be a Microsoft model"
X Link 2025-08-22T22:48Z [----] followers, 21.1K engagements
"Flash-2.5-Flash-Image (Nano-Banana) leads both the Text-to-Image and Image-Edit leaderboards on LmArena holding a big lead over Flux-1-kontext-Max in Image-Edit"
X Link 2025-08-26T14:26Z [----] followers, [----] engagements
"A potential new OpenAI model King-Kedra-0827 is being tested on WebdevArena. After the voting process the OpenAI logo appears beside the name and the model identifies itself as an OpenAI model OpenAI has tested a few models on WebdevArena before with Anonymous-Chatbot-0717 the most similar by naming scheme both names ending in a numeric suffix that likely marks the checkpoint date"
X Link 2025-08-30T06:20Z [----] followers, 31.6K engagements
"A new Image model "DH3" is currently being tested on the Artificial Analysis Image Arena From initial testing the model seems to be on par with Nano-Banana at Image editing According to @bdsqlsz there are two Chinese models one open-weights and one proprietary that are about to be announced and are comparable to Nano-Banana Hopefully this is the open-weights one It's a bit crazy there are two image models about to be announced that are comparable to the nano banana.😋 It's a bit crazy there are two image models about to be announced that are comparable to the nano banana.😋"
X Link 2025-09-05T07:46Z [----] followers, 26.4K engagements
"2 new mystery models "Sonoma Sky Alpha" and "Sonoma Dusk Alpha" will soon be dropped on Openrouter"
X Link 2025-09-05T20:29Z [----] followers, [----] engagements
"New Qwen-3-Next-80B-A3B model incoming "Built on this architecture we trained and open-sourced Qwen3-Next-80B-A3B 80B total parameters only 3B active achieving extreme sparsity and efficiency. Despite its ultra-efficiency it outperforms Qwen3-32B on downstream tasks while requiring less than 1/10 of the training cost. Moreover it delivers over 10x higher inference throughput than Qwen3-32B when handling contexts longer than 32K tokens" https://t.co/HbDdaZOO5L https://t.co/MtS1N60nqR https://t.co/HbDdaZOO5L https://t.co/MtS1N60nqR"
X Link 2025-09-09T14:40Z [----] followers, 85.2K engagements
"A new Google Gemini / Gemma model "Oceanstone" is being tested in LmArena"
X Link 2025-09-15T06:54Z [----] followers, 177.4K engagements
"Another new Google Gemini model "Oceanreef" is being tested in LmArena The model is likely related to the "Oceanstone" model which appeared [--] days ago"
X Link 2025-09-17T05:47Z [----] followers, 13.7K engagements
"Potential new Grok code model "Code-Supernova" is currently being tested in Cursor and Cline"
X Link 2025-09-20T08:35Z [----] followers, 16.6K engagements
"New Qwen3-Omni-7B model is soon to be released"
X Link 2025-09-20T18:21Z [----] followers, 49.6K engagements
"New Deepseek v3 version "Terminus" incoming"
X Link 2025-09-22T11:48Z [----] followers, 56.7K engagements
"Potential new Grok model "Rainbow" is currently being tested in WebDevArena"
X Link 2025-09-25T09:02Z [----] followers, [----] engagements
"New Checkpoints for Gemini [---] Flash and Flash-Lite This latest [---] Flash model comes with improvements in two key areas we heard consistent feedback on: - Better agentic tool use: We've improved how the model uses tools leading to better performance in more complex agentic and multi-step applications This model shows noticeable improvements on key agentic benchmarks including a 5% gain on SWE-Bench Verified compared to our last release (48.9% 54%) - More efficient: With thinking on the model is now significantly more cost-efficientachieving higher quality outputs while using fewer tokens"
X Link 2025-09-25T17:28Z [----] followers, 14.4K engagements
"New Image model "Lavender" is currently being tested on the Artificial Analysis Arena The model is likely related to the GPT-image-1 model possibly an update Prompt: A cartoonish 1940s detectives office rendered in a loose watercolor style: a gumshoe in a pinstriped suit fedora tilted low studies a case file beneath a single incandescent bulb. Smoke curls from his cigarette and old newspapers a rotary phone and a half-empty whiskey glass set the moody stage"
X Link 2025-10-03T15:08Z [----] followers, 21.6K engagements
"Gemini [--] Pro () Zenith (GPT-5-Checkpoint) - Doom Game This is probably the best result so far from a Gemini model for this prompt However it still lacks some of the functionality that Zenith had such as the minimap sound effects and enhanced health UI The enemies though look significantly better in what the Gemini model produced"
X Link 2025-10-05T13:21Z [----] followers, 22K engagements
"Gemini [--] "ecpt" Gemini [---] Pro - Space Invaders I used to get significantly better results from Gemini [---] Pro but its performance seems to have declined on this prompt The output shown in the video is the best I got after multiple tries while Gemini 3s result is from the first try"
X Link 2025-10-12T11:41Z [----] followers, 96.8K engagements
"Gemini [--] Pro related strings found on the Gemini site. The release should be really close now the strings dont lie "We've upgraded you from the previous model to [---] Pro our smartest model yet." the strings dont lie "We've upgraded you from the previous model to [---] Pro our smartest model yet.""
X Link 2025-10-15T19:14Z [----] followers, 20.2K engagements
"2 new Google Gemini models "Orionmist" & "Lithiumflow" are currently being tested in LmArena"
X Link 2025-10-19T18:18Z [----] followers, 54.6K engagements
"Gemini [--] "9d30" Lithiumflow Orionmist Gemini [---] Pro - 3D Voxel first gen starter Pokemon scene Lithiumflow and Orionmist seem slightly less consistent and overall weaker than some of the Gemini [--] checkpoints Ive tested on AI Studio Still they demonstrate a significant improvement over Gemini [---] Pro"
X Link 2025-10-24T10:13Z [----] followers, 19.7K engagements
"Oak Willow - Xbox Controller SVG These [--] mystery models are likely different GPT-5 checkpoints 🚨 Breaking News DesignArena [--] New Mystery Models willow cedar birch oak You can get this model superfast compared to Lmarena in designarena link in comment and also these models are from big lab of USA iykyk 😉 https://t.co/ha7e2Uedgs 🚨 Breaking News DesignArena [--] New Mystery Models willow cedar birch oak You can get this model superfast compared to Lmarena in designarena link in comment and also these models are from big lab of USA iykyk 😉 https://t.co/ha7e2Uedgs"
X Link 2025-11-01T11:16Z [----] followers, 28.8K engagements
"New GPT-5 checkpoints have been added to DesignArena The reasoning budget of the new models: Firefly: [--] Chrysalis: [--] Cicada: [--] Caterpillar: [---] The new checkpoints seem a bit better than the old ones from initial testing 🚨 Breaking News - DesignArena [--] New Mystery Models cicada caterpillar chrysalis firefly You can get this model superfast compared to Lmarena in designarena link in comment and also these models are from big lab of USA😉 https://t.co/dCh6DFgyTf 🚨 Breaking News - DesignArena [--] New Mystery Models cicada caterpillar chrysalis firefly You can get this model superfast compared"
X Link 2025-11-04T09:58Z [----] followers, 20.1K engagements
"The GPT-5 models that were being tested on DesignArena are likely GPT-5-1-Thinking with different reasoning budgets There are also rumors that GPT-5-1-Thinking will have a bigger context window and reduced API pricing LEAK: GPT-5-1 Thinking officially confirmed by OpenAI https://t.co/lLzvZcMVmT LEAK: GPT-5-1 Thinking officially confirmed by OpenAI https://t.co/lLzvZcMVmT"
X Link 2025-11-06T10:11Z [----] followers, 15.2K engagements
"Kimi-K2-Thinking beats GPT-5-Pro on Humanitys Last exam Kimi-K2-Thinking-Heavy achieves over 50% on Humanitys Last exam"
X Link 2025-11-06T16:05Z [----] followers, 72.3K engagements
"Grok Image "Mandarin"Nano-Banana - Last paragraph from "Faust" The new Grok Image model "Mandarin" seems really good text generation capabilities seem to be on par with Nano-Banana Impressive how quickly xAI is improving and iterating on their image and video generation models 🚨 New Grok Image Model on LM Arena: Mandarin Looks pretty solid so far Ill be testing it more soon. Its from xAI. We got some competition for NB2 https://t.co/9Cs1KZlzdj 🚨 New Grok Image Model on LM Arena: Mandarin Looks pretty solid so far Ill be testing it more soon. Its from xAI. We got some competition for NB2"
X Link 2025-11-11T10:09Z [----] followers, 32.4K engagements
"A new Google Gemini model "Riftrunner" is currently being tested on LmArena"
X Link 2025-11-12T08:22Z [----] followers, 34.3K engagements
"Gemini [--] appears to be rolling out now The Canvas feature in the mobile app seems to use the new Gemini [--] model The difference between Web and mobile for the 3D Pokemon voxel scene is huge It seems Gemini [--] is secretely rolling-out on the Gemini app using the Canvas feature Big difference in output quality between web (PC) and mobile versions Left: Web version Right: Mobile version https://t.co/bXPosvA65V It seems Gemini [--] is secretely rolling-out on the Gemini app using the Canvas feature Big difference in output quality between web (PC) and mobile versions Left: Web version Right: Mobile"
X Link 2025-11-13T08:08Z [----] followers, 118.1K engagements
Limited data mode. Full metrics available with subscription: lunarcrush.com/pricing