# ![@TeksEdge Avatar](https://lunarcrush.com/gi/w:26/cr:twitter::1682428283135336449.png) @TeksEdge David Hendrickson

David Hendrickson posts on X about ai, open ai, $googl, openclaw the most. They currently have [-----] followers and [---] posts still getting attention that total [-----] engagements in the last [--] hours.

### Engagements: [-----] [#](/creator/twitter::1682428283135336449/interactions)
![Engagements Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1682428283135336449/c:line/m:interactions.svg)

- [--] Week [-------] +6.30%
- [--] Month [-------] +381%
- [--] Months [-------] -55%
- [--] Year [---------] +1,027%

### Mentions: [--] [#](/creator/twitter::1682428283135336449/posts_active)
![Mentions Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1682428283135336449/c:line/m:posts_active.svg)

- [--] Month [---] +47%
- [--] Months [---] +51%
- [--] Year [-----] +207%

### Followers: [-----] [#](/creator/twitter::1682428283135336449/followers)
![Followers Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1682428283135336449/c:line/m:followers.svg)

- [--] Week [-----] +5.30%
- [--] Month [-----] +26%
- [--] Months [-----] +36%
- [--] Year [-----] +518%

### CreatorRank: [-------] [#](/creator/twitter::1682428283135336449/influencer_rank)
![CreatorRank Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1682428283135336449/c:line/m:influencer_rank.svg)

### Social Influence

**Social category influence**
[technology brands](/list/technology-brands)  48.72% [stocks](/list/stocks)  14.1% [finance](/list/finance)  6.41% [countries](/list/countries)  3.21% [social networks](/list/social-networks)  2.56% [currencies](/list/currencies)  1.92% [celebrities](/list/celebrities)  1.28% [automotive brands](/list/automotive-brands)  0.64% [financial services](/list/financial-services)  0.64%

**Social topic influence**
[ai](/topic/ai) 16.67%, [open ai](/topic/open-ai) #2072, [$googl](/topic/$googl) #3769, [openclaw](/topic/openclaw) #639, [anthropic](/topic/anthropic) #2528, [in the](/topic/in-the) 7.69%, [model](/topic/model) #3976, [ultra](/topic/ultra) #3879, [inference](/topic/inference) #435, [agentic](/topic/agentic) 3.85%

**Top accounts mentioned or mentioned by**
[@openai](/creator/undefined) [@alexfinn](/creator/undefined) [@openclaw](/creator/undefined) [@arena](/creator/undefined) [@anthropicai](/creator/undefined) [@zaiorg](/creator/undefined) [@steipete](/creator/undefined) [@openrouterai](/creator/undefined) [@openais](/creator/undefined) [@google](/creator/undefined) [@sama](/creator/undefined) [@googles](/creator/undefined) [@artificialanlys](/creator/undefined) [@minimaxai](/creator/undefined) [@matthewberman](/creator/undefined) [@iamemily2050](/creator/undefined) [@digitalix](/creator/undefined) [@latest](/creator/undefined) [@testingcatalog](/creator/undefined) [@kimmonismus](/creator/undefined)

**Top assets mentioned**
[Alphabet Inc Class A (GOOGL)](/topic/$googl) [Microsoft Corp. (MSFT)](/topic/microsoft) [Goldman Sachs (GS)](/topic/goldman-sachs) [Atlassian Corporation PLC (TEAM)](/topic/$team)
### Top Social Posts
Top posts by engagements in the last [--] hours

"@LinusEkenstam The price difference is stark and the 5x difference is extreme but cards have different capabilities. It's like comparing an AMD RX [----] (24GB) $700 to an Nvidia RTX 5090m (24GB) $2000"  
[X Link](https://x.com/TeksEdge/status/2016913791406264801)  2026-01-29T16:38Z [----] followers, [----] engagements


"If youre not already aware Clawdbot changed to Moltbot changed to OpenClaw. Then they got their social media platform Moltbook then created their own religion crustafarianism"  
[X Link](https://x.com/TeksEdge/status/2017266704045494646)  2026-01-30T16:00Z [----] followers, [----] engagements


"@karpathy Its finally dawning on the people inside the big AI labs that they're not going to be exempt. They lived under the comforting illusion that the labor-deflation their systems would unleash would fall on everyone else. Then Clawdbot showed up and broke the spell"  
[X Link](https://x.com/TeksEdge/status/2017360269182472584)  2026-01-30T22:12Z [----] followers, [----] engagements


"🦞 OpenClaw News: Huge OpenClaw update committed Feb [--] [----]. Just released. Security Updates I dug into the tickets and here is what I see. 👀 The post-rebrand momentum continues with a major focus on Agent Safety and Developer Control. ✨ New Additions 🛡 System Guardrails: 👀 Native prompt injection defense to keep agents secure. 👀 🧠 Session Injection: CreateAgentSessionOptions now accepts direct systemPrompt skills & contextFiles for total context control. Precision: Timestamps now baked into all Gateway messages. 🔧 Fixes & Polish 💬 Discord: Thread parent binding fixedno more lost"  
[X Link](https://x.com/TeksEdge/status/2018358790253670421)  2026-02-02T16:20Z [----] followers, [---] engagements


"❔ People ask What's The Difference Between Claude Code/Cowork and Codex and 🦞 OpenClaw / Clawdbot Benefits of Using Clawdbot (OpenClaw) Over Proprietary Alternatives Clawdbot (now known as OpenClaw) stands out as an open-source locally-run AI agent framework that prioritizes user control flexibility and proactivity. Here is a breakdown of key advantages over @Anthropic's Claude Code + Cowork ecosystem and @OpenAI's Codex based on their architectural and philosophical differences. Zero Cost and No Subscription Lock-In Clawdbot is completely free to use and self-host with no recurring fees or"  
[X Link](https://x.com/TeksEdge/status/2018447789831934403)  2026-02-02T22:13Z [----] followers, [---] engagements


"🦞 Clawdbot News: According to OpenRouter here are the top Clawdbot (OpenClaw) LLMs being used to power the brain of the bots. Keep in mind these are only the API calls managed through OpenRouter AND Kimi K2.5 is currently free. Gemini [--] Flash is taking the prize. Good to know that Kimi K2.5 is close behind and can be run locally at home (for those with 2xMacUltra's 512GB) Tip: You can use OpenRouter to select only free APIs and will do so automagically. [--]. Gemini [--] Flash Preview [--]. Kimi K2.5 [--]. Trinity Large Preview (free) [--]. Claude Sonnet [---] [--]. Claude Opus [---] (expensive $$$$$$) The top"  
[X Link](https://x.com/TeksEdge/status/2018715718158811553)  2026-02-03T15:58Z [----] followers, [---] engagements


"@sundarpichai Congratulations on the $400B quarterly revenue. Certainly deserved after many in the AI-X community thought you guys were finished following the ChatGPT moment. You turned the ship"  
[X Link](https://x.com/TeksEdge/status/2019157058772435297)  2026-02-04T21:12Z [----] followers, 14.1K engagements


"@mattshumer_ Matt if that's true then METR task times will officially cross the [--] hour human workday barrier which is maybe what Sam Altman was talking about the public is not prepared. 👀 😱"  
[X Link](https://x.com/TeksEdge/status/2019480281905013024)  2026-02-05T18:36Z [----] followers, [----] engagements


"@mitsuhiko Don't tell Linus. 🙂 I've Vibe Code a few operating systems for small hardware and microcontrollers. Do people realize this can be done Do they care"  
[X Link](https://x.com/TeksEdge/status/2019493465323884736)  2026-02-05T19:28Z [----] followers, [----] engagements


"The problem is that this trend began to rise before Codex [---] and 🦞 Clawdbot (OpenClaw). I think 20% AI written commits by the end of [----] is extremely conservative. 4% of GitHub public commits are being authored by Claude Code right now. At the current trajectory we believe that Claude Code will be 20%+ of all daily commits by the end of [----]. While you blinked AI consumed all of software development. Read more 👇 https://t.co/HzK4nbe2vy https://t.co/E1kIjfrNgk 4% of GitHub public commits are being authored by Claude Code right now. At the current trajectory we believe that Claude Code will"  
[X Link](https://x.com/TeksEdge/status/2019523630703734950)  2026-02-05T21:28Z [----] followers, [---] engagements


"📈Check out this tweet. Sergey has a new benchmark that he ran Opus [---] and Codex [---]. Codex won on quality and cost. 💰 A new benchmark run by Sergey just pitted Claude Opus [---] against OpenAI Codex [---] and the results show why pass rate charts are useless. Codex won not because it was "smarter" but because it was cheaper and required less cleanup. Here is the shift: Most AI benchmarks (like SWE-bench) are pass/fail with strict static rules totally disconnected from real work. This new approach is a live battle. It doesn't use a test file. It uses your actual codebase. You open a ticket"  
[X Link](https://x.com/TeksEdge/status/2019565094095712337)  2026-02-06T00:13Z [----] followers, [---] engagements


"@AlexFinn Awesome I cant wait to follow your experience. Running Kimi K2.5 at home with 🦞 Clawdbot is the black swan scenario that invalidates Big AIs plan for inference scale up in space and consumer lock in"  
[X Link](https://x.com/TeksEdge/status/2019583667560169646)  2026-02-06T01:27Z [----] followers, [---] engagements


"@Grummz Making too much money with hyperscalers. Irony would be Huawei selling video cards and DRAM into the US to meet demand"  
[X Link](https://x.com/TeksEdge/status/2019606312645169153)  2026-02-06T02:57Z [----] followers, [---] engagements


"Yes Vibe Coding is a thing and the value as a result is declining"  
[X Link](https://x.com/TeksEdge/status/2019638853196484960)  2026-02-06T05:06Z [----] followers, [---] engagements


"@sama I took me a minute that realize Codex [---] wasn't a version of the coding environment"  
[X Link](https://x.com/TeksEdge/status/2019814048158933111)  2026-02-06T16:42Z [----] followers, [----] engagements


"Here is where to merge the power of world modeling from @Google's Genie [--] and merge it with real-world applications. Waymo uses it to "think" while it drives. Amazing. You want your car to be smart but not "human" smart. It might start driving itself around when it is bored. Were excited to introduce the Waymo World Modela frontier generative mode for large-scale hyper-realistic autonomous driving simulation built on @GoogleDeepMinds Genie [--]. By simulating the impossible we proactively prepare the Waymo Driver for some of the most rare and https://t.co/Pl80OMDqLC Were excited to introduce the"  
[X Link](https://x.com/TeksEdge/status/2019865192679747605)  2026-02-06T20:06Z [----] followers, [--] engagements


"📢 What's the Deal with @heroku☠ The Post-Vibe Coding Era has Arrived. Lots of chatter on X today about Heroku. It is shocking that the era of Heroku as we knew it is ending but it is a sign of the times. ⚠ Vanilla AI Agents can build and deploy websites and backends in general with no need for a special platform. Heroku launched in [----] was one of the pioneers of Platform as a Service (PaaS). It made deploying and running web applications dramatically easier for developers. You could build an app locally push it with a simple git push heroku main and Heroku handles servers scaling databases"  
[X Link](https://x.com/TeksEdge/status/2019889731606311112)  2026-02-06T21:43Z [----] followers, [---] engagements


"Betting markets are saying @AnthropicAI will dominate with the best model thsi month. I don't know. The jury is still out on Codex [---] versus Opus [---]. BREAKING: Anthropic dethrones Google as the #1 projected AI this month. https://t.co/Ft1amWRSlV BREAKING: Anthropic dethrones Google as the #1 projected AI this month. https://t.co/Ft1amWRSlV"  
[X Link](https://x.com/TeksEdge/status/2019891832340246775)  2026-02-06T21:51Z [----] followers, [---] engagements


"@testingcatalog Codex or Claude Code for Grok Is Thai included for X Premium Plus users too"  
[X Link](https://x.com/TeksEdge/status/2019916391265234955)  2026-02-06T23:29Z [----] followers, [---] engagements


"Ever wonder why @Microsoft has been so absent from the LLM battles Google Anthropic xAI Meta and OpenAI are engaged in It concerns Microsoft's ownership of OpenAI's IP. Remember the partnership ends when AGI is declared but others on X have also stated; "Microsoft will have a copy of it (GPT-AGI) and will need to further develop it separately from OpenAI." This is why Microsoft isn't in the current LLM game: they don't need to jump in until OpenAI achieves AGI. Then Microsoft jumps into the pool. Keep in mind that Microsoft has the largest data center on the planet."  
[X Link](https://x.com/TeksEdge/status/2020010689855909989)  2026-02-07T05:44Z [----] followers, [---] engagements


"@vitrupo Big bet is that everyone will need (and pay for) Hyperscaler tokens. You can generate tokens at home for free"  
[X Link](https://x.com/TeksEdge/status/2020011808778748335)  2026-02-07T05:48Z [----] followers, [---] engagements


"If Goldman Sachs builds agents to replace its workers using Anthropic's Claude then ultimately only a skeleton crew of (rich) humans will run it. What prevents Anthropic from turning off access to Goldman Sachs and simply offering the same Consulting Agents to Goldman customers"  
[X Link](https://x.com/TeksEdge/status/2020120363263267075)  2026-02-07T13:00Z [----] followers, [---] engagements


"Anthropic please don't destroy the books. You can send them my way. 🙂"  
[X Link](https://x.com/TeksEdge/status/2020120364601254268)  2026-02-07T13:00Z [----] followers, [--] engagements


"🤔 You are a potential investor in Atlassian a pure SaaS company. You attend an investor meeting where someone in the audience asks why we should invest in Atlassian and the co-founder answers with this. AI at Atlassian grows revenue because: Coding is only about 2030% of a developers job so even big gains in code generation translate to a more modest 1015% productivity gain across the whole engineering org. When teams become more productive they rarely reduce scope but instead they add more features products and experiments to the roadmap to keep them working. This leads to more applications"  
[X Link](https://x.com/TeksEdge/status/2020135460006150624)  2026-02-07T14:00Z [----] followers, [---] engagements


"🦞 New Clawdbot Update: New Release now supporting Opus [---] and Codex GPT-5.3. Massive security hardening meets new models AI (Opus [---] & GPT-5.3). If you run a public gateway upgrade immediately. 🤖 1/ AI & MODELS Opus [---] & GPT-5.3: Core runner updated for the latest reasoning/coding checkpoints. 🔒 2/ SECURITY (CRITICAL) Attachments are now strictly locked to allowed directories to prevent path traversal. (Thanks @victormier) Overrides now require explicit credentials to prevent leakage vectors. WhatsApp Gating (#8768): whatsapp_login is restricted to "owner" contexts only. Metadata is"  
[X Link](https://x.com/TeksEdge/status/2020199839188083055)  2026-02-07T18:15Z [----] followers, [---] engagements


"Can't help to think that Clawdbot (OpenClaw) is the reason why Kimi K2.5 continues to be the top used model on OpenRouter"  
[X Link](https://x.com/TeksEdge/status/2020202222441885945)  2026-02-07T18:25Z [----] followers, [----] engagements


"The leaderboards are announcing updates and Claude [---] is sitting on top of most. Why didn't @OpenAI add Codex [---] to the mix. If AI-X community is confused about GPT naming surely the public is too. Is Codex GPT-5.3 a separate model BREAKING: Claude Opus [---] and its thinking variant have taken #1 and #2 on Design Arena This means [--] of the top [--] models on Design Arena belong to Anthropic Congrats to the @AnthropicAI team for this achievement https://t.co/XFFz1Suioz BREAKING: Claude Opus [---] and its thinking variant have taken #1 and #2 on Design Arena This means [--] of the top [--] models on"  
[X Link](https://x.com/TeksEdge/status/2020218224089186618)  2026-02-07T19:28Z [----] followers, [---] engagements


"@kimmonismus Interesting Will @AnthropicAI be able to sustain their business with this insane pricing Theyre putting pressure on the Viber Coders. We will be watching closely. Kimi K2.5 inferencing locally is more appealing now"  
[X Link](https://x.com/TeksEdge/status/2020233934983074030)  2026-02-07T20:31Z [----] followers, [----] engagements


"Interesting How does the average person verify this assertion - Answer: We cant. This is the AGI/ASI boundary. What is the difference between false and incomprehensible History just happened. An AI system has autonomously solved a long standing open conjecture in pure mathematics (Fels open conjecture on syzygies of numerical semigroups) building a complete formal proof in Lean with zero human input. This is the first time in history an History just happened. An AI system has autonomously solved a long standing open conjecture in pure mathematics (Fels open conjecture on syzygies of numerical"  
[X Link](https://x.com/TeksEdge/status/2020237756677738573)  2026-02-07T20:46Z [----] followers, [---] engagements


"@AnonWolfoRagnar @kimmonismus @AnthropicAI True way out of my range BUT this is still a phenomenal shift. A single H100 SXM5 costs $30K with only 80GB VRAM (can't run Kimi K2.5). So running Kimi K2.5 at home for $20K is a game changer"  
[X Link](https://x.com/TeksEdge/status/2020257515179766117)  2026-02-07T22:05Z [----] followers, [---] engagements


"@thorstenball @steipete Problem is Anthropic and OpenAI will continue to raise their API pricing making inferencing with top models more expensive. This will drive token hungry bots like Clawdbot to local token generation"  
[X Link](https://x.com/TeksEdge/status/2020291595485213042)  2026-02-08T00:20Z [----] followers, [---] engagements


"Will Qwen3.5 dethrone Kimi K2.5 as the Clawdbot (open source) brain of choice Can't help to think that Clawdbot (OpenClaw) is the reason why Kimi K2.5 continues to be the top used model on OpenRouter. https://t.co/9cUuPWHOM0 Can't help to think that Clawdbot (OpenClaw) is the reason why Kimi K2.5 continues to be the top used model on OpenRouter. https://t.co/9cUuPWHOM0"  
[X Link](https://x.com/TeksEdge/status/2020548912776634860)  2026-02-08T17:22Z [----] followers, [---] engagements


"@AlexFinn I take it things are going well With Anthropic raising the cost of output tokens of [---] one can't help to dream of an at-home setup where you get unlimited tokens that are 90% as good as Opus 4.6"  
[X Link](https://x.com/TeksEdge/status/2020569127438647521)  2026-02-08T18:43Z [----] followers, [---] engagements


"@chetaslua @Alibaba_Qwen @arena Can't wait. Also will be curious to see parameter count. Will they build a 600B model we can run at home like Kimi K2.5"  
[X Link](https://x.com/TeksEdge/status/2020569612887162971)  2026-02-08T18:45Z [----] followers, [---] engagements


"@realBigBrainAI He says this a lot. Apparently he hasn't been successful convincing Meta xAI OpenAI Anthropic Microsoft Google "  
[X Link](https://x.com/TeksEdge/status/2020610443430269249)  2026-02-08T21:27Z [----] followers, [----] engagements


"Well today is Ad-pocolypse for @OpenAI. The real question is how much does it look like @perplexity_ai or @Google Gemini's ads"  
[X Link](https://x.com/TeksEdge/status/2020896731349880835)  2026-02-09T16:25Z [----] followers, [---] engagements


"More evidence we are at an inflection point. 🦞 OpenClaw (Clawdbot) versus n8n a popular node-based workflow automation tool. who's next https://t.co/rrgul7UiQh who's next https://t.co/rrgul7UiQh"  
[X Link](https://x.com/TeksEdge/status/2020910067625943472)  2026-02-09T17:18Z [----] followers, [---] engagements


"🦞 Clawdbot News: A new competitor is in town to challenge 🦞 OpenClaw's dominance. It's available running on pico-hardware. Vibe Porting is real and easy. #PicoClaw: AI built the code in hours matching #OpenClaws core features with only 1% code and 1% memory Ditch your Mac Mininow you can run a full AI assistant on $10 RISCV hardware with 10MB RAM If it runs Linux it can now be your personal AI Agent https://t.co/OF6rG9kgev https://t.co/iXAczSl7VN #PicoClaw: AI built the code in hours matching #OpenClaws core features with only 1% code and 1% memory Ditch your Mac Mininow you can run a full"  
[X Link](https://x.com/TeksEdge/status/2020921690210046093)  2026-02-09T18:04Z [----] followers, [---] engagements


"@googleaidevs Is this fine-tuning on HuggingFace free In other words can we fine tun on any data we want free of charge or do we have to rent GPUs on HF to do this"  
[X Link](https://x.com/TeksEdge/status/2020940759709700167)  2026-02-09T19:19Z [----] followers, [---] engagements


"🦞 Clawdbot News: New update just dropped The latest update is rolling out now bringing critical security defenses and next-gen model support to your local Clawdbot stack. : Here are the key changes in v2026.2.9 📱 Easier setup + early mobile support: Theres now an alpha iOS node app and onboarding for setup code. This release also adds device pairing and phone control plugins including Telegram commands such as/pair and /pair approve as well as iOS/Android node controls. 🖥 More control from the Web UI: The Gateway now supports Web UI actions for creating updating and deleting agents"  
[X Link](https://x.com/TeksEdge/status/2020981475387248881)  2026-02-09T22:01Z [----] followers, [---] engagements


"@AlexFinn Today's Clawdbot software update improves my Clawdbots' performance a little. Highly recommend updating"  
[X Link](https://x.com/TeksEdge/status/2021010444526456905)  2026-02-09T23:56Z [----] followers, [--] engagements


"🙄Good grief. Algorithms versus LLMs boils down to Deterministic versus Probablistic. Humans will never hand over control until we approach "Deterministic" w/"Probability" Six Sigma. 😠 These hallucination rates and HalluHard benchmark will not cut it. The hallucination rate gap between proprietary and open-weight models remains very substantial on HalluHard (GPT-5.2-thinking = 53.8% DeepSeek-Chat = 75.8%). Best open-weight models appear to be "benchmaxxed" and new benchmarks like ours reflect this. https://t.co/n4jatNMhCI https://t.co/c9m5EotoVi The hallucination rate gap between proprietary"  
[X Link](https://x.com/TeksEdge/status/2021012372136263909)  2026-02-10T00:04Z [----] followers, [---] engagements


"@FarzadClaw Well done. It captures things nicely. Normies still struggle with the realities of the technology and CIOs struggle with a high degree of Dunning-Kruger. Digital Moats are gone and business expertise asymmetries are also a thing of the past"  
[X Link](https://x.com/TeksEdge/status/2021090879453397485)  2026-02-10T05:16Z [----] followers, [---] engagements


"Wow I didnt realize @ltx_model has a new open-source TTV competitor. Its called Seedance [--]. I cant wait to test side by side in @ComfyUI. Seeddance [---] ai 'Figure [--] Figure [--] ' https://t.co/eEF8F1sJhK Seeddance [---] ai 'Figure [--] Figure [--] ' https://t.co/eEF8F1sJhK"  
[X Link](https://x.com/TeksEdge/status/2021244663252779025)  2026-02-10T15:27Z [----] followers, [---] engagements


"Indeed benchmarks aren't everything but I think @OpenAI's models need a refresh. They currently rank at the bottom of the @arena leaderboards"  
[X Link](https://x.com/TeksEdge/status/2021262999688774037)  2026-02-10T16:40Z [----] followers, [---] engagements


"I bet by the end of this year we will see bargain basement prices on PCs as a growing backlog of DRAM inventory sits in warehouses collecting dust waiting for AI data centers to go online"  
[X Link](https://x.com/TeksEdge/status/2021265736820244518)  2026-02-10T16:51Z [----] followers, [---] engagements


"@btibor91 No doubt quickly responding to the 🦞 Clawdbot Skills or @AnthropicAI's Cowrok Plugins"  
[X Link](https://x.com/TeksEdge/status/2021267333189189806)  2026-02-10T16:57Z [----] followers, [--] engagements


"@iam_smx Drinking Elon Kool-Aid but not wrong. Future of tech is hardware and power"  
[X Link](https://x.com/TeksEdge/status/2021267977207677306)  2026-02-10T17:00Z [----] followers, [--] engagements


"OpenAI is releasing fast and coming up to speed to match 🦞 Clawdbot and Cowork core functionality. They just announced memory compaction skills templates and domain access control. We're introducing a new set of primitives in the Responses API for long-running agentic work on computers. Server-side compaction Enable multi-hour agent runs without hitting context limits. Containers with networking Give OpenAI-hosted containers controlled internet https://t.co/6wFxK0YbAa We're introducing a new set of primitives in the Responses API for long-running agentic work on computers. Server-side"  
[X Link](https://x.com/TeksEdge/status/2021295537451143252)  2026-02-10T18:49Z [----] followers, [---] engagements


"So Salesforce reduced heacount by [----] this month Seattle is taking a jobs beating"  
[X Link](https://x.com/TeksEdge/status/2021302800630657133)  2026-02-10T19:18Z [----] followers, [---] engagements


"A new company is jumping into the Vibe Coding business. Are they too late We seem to be moving into the Vibe Agent Coding era"  
[X Link](https://x.com/TeksEdge/status/2021304314858279210)  2026-02-10T19:24Z [----] followers, [---] engagements


"@Scobleizer @openclaw @Team9_ai It even kindled a fire in OpenAI to ship more features for Codex"  
[X Link](https://x.com/TeksEdge/status/2021310825344983131)  2026-02-10T19:50Z [----] followers, [--] engagements


"@steipete The AI spark the EU needs"  
[X Link](https://x.com/TeksEdge/status/2021310967120810112)  2026-02-10T19:51Z [----] followers, [--] engagements


"@OpenAI is that a good thing Why not Codex 5.3"  
[X Link](https://x.com/TeksEdge/status/2021311432235929613)  2026-02-10T19:52Z [----] followers, [---] engagements


"@AlexFinn I'm hearing good things about beta GLM-5 (code named Poly Alpha)"  
[X Link](https://x.com/TeksEdge/status/2021343302134661367)  2026-02-10T21:59Z [----] followers, [---] engagements


"@apples_jimmy What about LTX-2 Way better"  
[X Link](https://x.com/TeksEdge/status/2021352062941106481)  2026-02-10T22:34Z [----] followers, [----] engagements


"👀 This is a Must Read (link in the ALT.) 📰 This article is unusually honest about the gap between the story and the scoreboard. Economists cite past innovations as proof that new tech creates new jobs but the piece also notes that the data still dont show AI displacing jobs in a measurable way and nobody has a real plan if that changes quickly. Its been 3+ years since ChatGPT. Wheres the measurable upside 💡Here are the authors' main points; The author says economists looking at todays data mostly see no big effect yet (📉 I also have not been able to find a strong correlation in the data)"  
[X Link](https://x.com/TeksEdge/status/2021376758986178830)  2026-02-11T00:12Z [----] followers, [---] engagements


"@craigzLiszt To be honest most didnt really know how it worked before AI but point taken"  
[X Link](https://x.com/TeksEdge/status/2021455928198046039)  2026-02-11T05:27Z [----] followers, [--] engagements


"GPT-5.3 is immanent and is eagerly anticipated as current models are aging poorly on LM arena. Indeed benchmarks aren't everything but I think @OpenAI's models need a refresh. They currently rank at the bottom of the @arena leaderboards. https://t.co/04nxayTF97 Indeed benchmarks aren't everything but I think @OpenAI's models need a refresh. They currently rank at the bottom of the @arena leaderboards. https://t.co/04nxayTF97"  
[X Link](https://x.com/TeksEdge/status/2021469247495602623)  2026-02-11T06:20Z [----] followers, [---] engagements


"@koltregaskes But this take really follows the reliability vector. Humans won't voluntarily give up their agency until AI/Robots achieve 99.999% trustworthiness (Waymo Effect) BUT corporations are a totally different story"  
[X Link](https://x.com/TeksEdge/status/2021608154581147888)  2026-02-11T15:31Z [----] followers, [--] engagements


"@sama It goes to show the appetite for Codex in a world full of coding agents/frameworks/IDEs. The world is changing fast"  
[X Link](https://x.com/TeksEdge/status/2021609888678359219)  2026-02-11T15:38Z [----] followers, [---] engagements


"Opus [---] is leading in another benchmark. Currently Opus [---] and GPT-5.2 are the best closed source models with Kimi K2.5 best open source. Google Gemini needs and update and GLM-5 (releasing today) is going to be an open source beast We have done a cross-cutting independent benchmark of Claude Opus [---] on the OpenHands Index and it's the new top model for software engineering An improvement over [---] and GPT [---] Codex in both accuracy cost and time-to-resolution. https://t.co/waWW1La1jT We have done a cross-cutting independent benchmark of Claude Opus [---] on the OpenHands Index and it's the"  
[X Link](https://x.com/TeksEdge/status/2021611741147484200)  2026-02-11T15:46Z [----] followers, [----] engagements


"@ForHumanityPod The Super Bowl theory is that heavy tech advertising precedes a crash. [----] DotCom [----] Crypto [----] AI"  
[X Link](https://x.com/TeksEdge/status/2021623104666321392)  2026-02-11T16:31Z [----] followers, [--] engagements


"Can't argue with Free. New features migrating down to the free plans in Claude. We're bringing some of Claudes most-used features to the free plan. File creation connectors and skills are all now available without a subscription. https://t.co/6EjrwLTWVQ We're bringing some of Claudes most-used features to the free plan. File creation connectors and skills are all now available without a subscription. https://t.co/6EjrwLTWVQ"  
[X Link](https://x.com/TeksEdge/status/2021632219098558564)  2026-02-11T17:07Z [----] followers, [---] engagements


"Here are more benchmarks for GLM-5. The vending benchmark is of particular interest $$$$. This model can run locally with all the privacy and protection of a home network allowing users to inference at no cost. 3xMacUltras w/512GB is expensive but worth for some Clawdbots"  
[X Link](https://x.com/TeksEdge/status/2021640643513303228)  2026-02-11T17:41Z [----] followers, [---] engagements


"🚨 Wow GLM-5 is killing it on @ArtificialAnlys. The test results are now available. If you haven't tried it for free link in ALT. Otherwise check it out on @OpenRouterAI where some US-based providers are charging only $2.6/1M & better than Gemini [--] Pro. 🦞 Clawdbot powerup 💪 GLM-5 is the new leading open weights model GLM-5 leads the Artificial Analysis Intelligence Index amongst open weights models and makes large gains over GLM-4.7 in GDPval-AA our agentic benchmark focused on economically valuable work tasks GLM-5 is @Zai_org's first new https://t.co/IiK2GRptFL GLM-5 is the new leading"  
[X Link](https://x.com/TeksEdge/status/2021685258362695999)  2026-02-11T20:38Z [----] followers, [---] engagements


"@Zai_org Great job. Token costs just jumped too"  
[X Link](https://x.com/TeksEdge/status/2021686510660264403)  2026-02-11T20:43Z [----] followers, [----] engagements


"This is the pain of being in the inference business. GLM-5 just released and some benchmarks are showing better performance than @Google's Gemini [--] Pro AND Gemini [--] Flash and as good as Opus [---]. This is downward price pressure on both @Google @OpenAI and @AnthropicAI. Currently you can get GLM for $3.2/1M output tokens. https://twitter.com/i/web/status/2021690715022258359 https://twitter.com/i/web/status/2021690715022258359"  
[X Link](https://x.com/TeksEdge/status/2021690715022258359)  2026-02-11T21:00Z [----] followers, [---] engagements


"👀GLM-5 takes the open-source crown on the @arena. In head-to-head human-judged competition GLM-5 ranks #11 just behind GPT-5.1 and Gemini [--] Flash. Not too shabby Most importantly it ranks nearly [---] points better than Kimi K2.5. 🦞 @openclaw (Clawdbots) has a new brain 🧠 GLM-5 from @Zai_org just climbed to #1 among open models in Text Arena #1 open model on par with claude-sonnet-4.5 & gpt-5.1-high #11 overall; scoring [----] +11pts over GLM-4.7 Test it out in the Code Arena and keep voting well see how GLM-5 performs for agentic coding https://t.co/MajenrS0Qz GLM-5 from @Zai_org just climbed"  
[X Link](https://x.com/TeksEdge/status/2021739054380036291)  2026-02-12T00:12Z [----] followers, [--] engagements


"👀GLM-5 takes the open-source crown on the @arena. In head-to-head human-judged competition GLM-5 ranks #11 just behind GPT-5.1 and Gemini [--] Flash. Not too shabby Note that it only beats Kimi K2.5 by [--] points. 🦞 @openclaw (Clawdbots) should keep its Kimi K2.5 brain 🧠 GLM-5 from @Zai_org just climbed to #1 among open models in Text Arena #1 open model on par with claude-sonnet-4.5 & gpt-5.1-high #11 overall; scoring [----] +11pts over GLM-4.7 Test it out in the Code Arena and keep voting well see how GLM-5 performs for agentic coding https://t.co/MajenrS0Qz GLM-5 from @Zai_org just climbed to"  
[X Link](https://x.com/TeksEdge/status/2021741132334710796)  2026-02-12T00:20Z [----] followers, [---] engagements


"Opus [---] is currently king of the LLMs. Rumors of Gemini [---] and GPT-5.3 are swirling and we could see them this week. Opus [---] is leading in another benchmark. Currently Opus [---] and GPT-5.2 are the best closed source models with Kimi K2.5 best open source. Google Gemini needs and update and GLM-5 (releasing today) is going to be an open source beast https://t.co/4ouP4Q29VD Opus [---] is leading in another benchmark. Currently Opus [---] and GPT-5.2 are the best closed source models with Kimi K2.5 best open source. Google Gemini needs and update and GLM-5 (releasing today) is going to be an open"  
[X Link](https://x.com/TeksEdge/status/2021768903492542841)  2026-02-12T02:10Z [----] followers, [---] engagements


"@lgaa201 @OpenRouterAI Kimi k2.5"  
[X Link](https://x.com/TeksEdge/status/2021787422796001356)  2026-02-12T03:24Z [----] followers, [--] engagements


"@jasonkwon [-----] hardware engineers vibe coding what could go wrong 😁"  
[X Link](https://x.com/TeksEdge/status/2021795790667264073)  2026-02-12T03:57Z [----] followers, [--] engagements


"@ml_angelopoulos @Zai_org I still like Kimi K2.5. $1/1M tokens cheaper on OpenRouter than GLM-5 and only [--] points lower in benchmark score on @arena"  
[X Link](https://x.com/TeksEdge/status/2021807943931638110)  2026-02-12T04:45Z [----] followers, [--] engagements


"@r0ck3t23 Hes not wrong. AI can write in machine code for target architecture. Humans couldnt follow without translating every hex to opcodes"  
[X Link](https://x.com/TeksEdge/status/2021808749393199434)  2026-02-12T04:49Z [----] followers, [---] engagements


"@distributionat Havent seen this story before. Did Elon find out some employees scheming to build a new startup and just cut 50% Or just because he could"  
[X Link](https://x.com/TeksEdge/status/2021827499379015920)  2026-02-12T06:03Z [----] followers, [----] engagements


"I really think NotebookLM was a sleeper hit for Google"  
[X Link](https://x.com/TeksEdge/status/2021973529772892246)  2026-02-12T15:43Z [----] followers, [---] engagements


"Anthropic is announcing a $20M investment in a lobbying group dedicated to advocating for AI and its regulation. IMHO this is not a good thing especially now. Big AI has established itself and is now fighting off open-source models from China available to all for free. AI is being adopted faster than any technology in history. The window to get policy right is closing. Today were contributing $20m to Public First Action a new bipartisan org that will mobilize people and politicians who understand whats at stake. https://t.co/y9ZQNQ1Q8W AI is being adopted faster than any technology in"  
[X Link](https://x.com/TeksEdge/status/2021975302227079377)  2026-02-12T15:50Z [----] followers, [---] engagements


"🚨 Big Hardware News: 👀 Samsung just shipped the first HBM4 DRAM. This is the new high-bandwidth memory at [--] Gbps nearly 3x the previous generation. This powers the @nvidia Rubin GPU architecture. @nvidia will get 30% of its DRAM from Samsung. https://twitter.com/i/web/status/2021980191309144332 https://twitter.com/i/web/status/2021980191309144332"  
[X Link](https://x.com/TeksEdge/status/2021980191309144332)  2026-02-12T16:10Z [----] followers, [--] engagements


"🚨 New Model Alert: @MiniMax_AI officially announced M2.5 It beats Gemini [--] Pro on SWE-Bench Pro SWE-Bench Verified VIBE-Pro and BrowseComp 👀 It even beats Opus [---] at Multi-SWE-Bench and BFCL multi-turn. 💵💵 Most importantly at $1 per hour with [---] tps infinite scaling of long-horizon agents is now economically possible Introducing M2.5 an open-source frontier model designed for real-world productivity. - SOTA performance at coding (SWE-Bench Verified 80.2%) search (BrowseComp 76.3%) agentic tool-calling (BFCL 76.8%) & office work. - Optimized for efficient execution 37% faster at complex"  
[X Link](https://x.com/TeksEdge/status/2021983172112851200)  2026-02-12T16:22Z [----] followers, [---] engagements


"🦞 Clawdbot (OpenClaw) News: @MiniMax_AI just released MiniMax [---]. This is an open-source model with some amazing programming benchmark scores outperforming Gemini [--] Pro and in some cases Opus [---] This is the perfect model for home inferencing running on [--] x Mac Ultras"  
[X Link](https://x.com/TeksEdge/status/2021984983003672756)  2026-02-12T16:29Z [----] followers, [---] engagements


"While my 🦞 Clawdbots were disappointed 😢 GLM-5 was out of reach for home inferencing the newly released MiniMax2.5 just dropped and it is a coding monster and better yet can run locally on two expensive Mac Ultras 🙂🤖. Free Free Tokens 🪙🪙🪙 The official announcement just dropped on GLM-5. Its a big one @ 1.5TB so those looking to inference locally you'll need a quantized version. If you still want something close keep your Clawdbots on Kimi K2.5 with the 2xMacUltras. https://t.co/CMA5xeCT78 The official announcement just dropped on GLM-5. Its a big one @ 1.5TB so those looking to"  
[X Link](https://x.com/TeksEdge/status/2021985721872265256)  2026-02-12T16:32Z [----] followers, [---] engagements


"@MatthewBerman Good point. The programming benchmarks look really good but (benchmaxxing) What are you using locally Kimi K2.5 Something smaller Qwen3 variant"  
[X Link](https://x.com/TeksEdge/status/2021987773205668042)  2026-02-12T16:40Z [----] followers, [--] engagements


"@MatthewBerman What do you think of myclaw dot ai I think a cloud-based Clawdbot defeats the purpose"  
[X Link](https://x.com/TeksEdge/status/2021989647518462157)  2026-02-12T16:47Z [----] followers, [--] engagements


""Sir MiniMax just dropped MiniMax M2.5 and it's on par with Opus [---] on some programming benchmarks while being 20x cheaper""  
[X Link](https://x.com/TeksEdge/status/2021991881450959334)  2026-02-12T16:56Z [----] followers, [---] engagements


"@AlexFinn MiniMax [---] just dropped and it's better than Opus [---] in some benchmarks and it might even be able to run on a maxed-out Mac Ultra"  
[X Link](https://x.com/TeksEdge/status/2021994449455526277)  2026-02-12T17:06Z [----] followers, [---] engagements


"GPT-4o is retiring tomorrow but similar performing models can now run entirely on phones. Tomorrow at 10am PT legacy models (GPT-5 GPT-4o GPT-4.1 GPT-4.1 mini and OpenAI o4-mini) will be deprecated in ChatGPT. https://t.co/RJioBsLY6D Tomorrow at 10am PT legacy models (GPT-5 GPT-4o GPT-4.1 GPT-4.1 mini and OpenAI o4-mini) will be deprecated in ChatGPT. https://t.co/RJioBsLY6D"  
[X Link](https://x.com/TeksEdge/status/2021994804549497298)  2026-02-12T17:08Z [----] followers, [---] engagements


"@steipete Killer. GLM-5 now supported Next include MiniMax [---] 🚀 Its better than Opus [---] as some coding benchmarks. 🤯"  
[X Link](https://x.com/TeksEdge/status/2022014678554259593)  2026-02-12T18:27Z [----] followers, [---] engagements


"@Karmedge Reasoning-gate but then quickly O1 was released and we all forgot about it"  
[X Link](https://x.com/TeksEdge/status/2022044557542863062)  2026-02-12T20:26Z [----] followers, [---] engagements


"@AI_4_Healthcare AI-Twitter already knows this and skeptics are gradually converting. Maybe its just reflective therapy I had this moment the day GPT-4 came out [--] years ago and wrote a book to get over it"  
[X Link](https://x.com/TeksEdge/status/2022336004225634583)  2026-02-13T15:44Z [----] followers, [--] engagements


"@AlexFinn Can't wait to hear your experience. What model are you going to run"  
[X Link](https://x.com/TeksEdge/status/2019580492836905227)  2026-02-06T01:14Z [----] followers, [---] engagements


"Good news on the GLM5 front it appears we might just be able to get this model running on three maxed-out Mac Ultras. Hopefully its performance justifies the added Mac Ultra over the Kimi K2.5 model"  
[X Link](https://x.com/TeksEdge/status/2020902697541812300)  2026-02-09T16:48Z [----] followers, [---] engagements


"Investment commentators saying AI-generated software will never eclipse or displace the incumbent software industry miss the point. Software is a human tool. In the near future our Bots won't use incumbent software APIs A2A MCP @profgalloway"  
[X Link](https://x.com/TeksEdge/status/2021343635036577917)  2026-02-10T22:00Z [----] followers, [---] engagements


"According to @EpochAIResearch who studied the capabilities of modern robots we have a ways to go before we get our robot butlers"  
[X Link](https://x.com/TeksEdge/status/2021803039490023434)  2026-02-12T04:26Z [----] followers, [--] engagements


"Something is brewing with Codex at @OpenAI. We've seen Codex-GPT-5.3 but could we be getting more agentic features in Codex Some things we make are purely meant to bring joy others will profoundly affect how you get things done. We are getting ready to launch something that combines both and the team is so excited to share it soon. ✨ Some things we make are purely meant to bring joy others will profoundly affect how you get things done. We are getting ready to launch something that combines both and the team is so excited to share it soon. ✨"  
[X Link](https://x.com/TeksEdge/status/2021974694023532564)  2026-02-12T15:48Z [----] followers, 11.8K engagements


"I think @Google's new Gemini [--] Deep Think model just about maxed out the ARC-AGI-2 benchmark. Now to ARC-AGI-3. An updated & faster Gemini [--] Deep Think is taking off 🚀 Our smartest mode to date PhD-level reasoning to the most rigorous STEM challenges (models' gotta think harder). Gold medal-level results on Physics & Chemistry Olympiads. 🧪💻 Full details: https://t.co/9XBQwHSCYW https://t.co/aOlbjI8RKo An updated & faster Gemini [--] Deep Think is taking off 🚀 Our smartest mode to date PhD-level reasoning to the most rigorous STEM challenges (models' gotta think harder). Gold medal-level"  
[X Link](https://x.com/TeksEdge/status/2021986898248618282)  2026-02-12T16:36Z [----] followers, [----] engagements


"@sama Ah so GPT-5.3-Codex-Spark is not really a model but a high-speed inference version. I guess longer think time (shorter human time) will improve performance"  
[X Link](https://x.com/TeksEdge/status/2022013330555277610)  2026-02-12T18:21Z [----] followers, [---] engagements


"If you're still not sure Chinese-based open-source models have not caught up take a look at this chart. Indeed they have "mostly" caught up. MiniMax [---] or Kimi K2.5 or GLM-5. just getting started. our next-gen MiniMax-Mx in training rn excited to see how far we can push it also this is my favorite chart. as an engineer who's been in the trenches the whole way watching our oss model climb this fast hits different. closing in on SOTA and already https://t.co/g6CQumqDoo just getting started. our next-gen MiniMax-Mx in training rn excited to see how far we can push it also this is my favorite"  
[X Link](https://x.com/TeksEdge/status/2022013709225496854)  2026-02-12T18:23Z [----] followers, [---] engagements


"MiniMax [---] put up a good fight but couldn't beat GLM-5 or Kimi K2.5. Still a solid model and at $1/1M output tokens a great value. BREAKING: MiniMax-M2.5 has taken #8 on Design Arena MiniMax-M2.5 is a noticeable improvement improving 30+ Elo points from its predecessor MiniMax-M2.1. Open-weights models are setting a new standard for real-world design tasks with competitive pricing Huge congratulations to https://t.co/GtSqSXYlLI BREAKING: MiniMax-M2.5 has taken #8 on Design Arena MiniMax-M2.5 is a noticeable improvement improving 30+ Elo points from its predecessor MiniMax-M2.1. Open-weights"  
[X Link](https://x.com/TeksEdge/status/2022016747856179402)  2026-02-12T18:35Z [----] followers, [---] engagements


"Just amazing performance in this benchmark by @OpenHandsDev. You can see MiniMax [---] beating GPT-5.2 and Gemini [--] Flash. It ranks #4 on this benchmark"  
[X Link](https://x.com/TeksEdge/status/2022017598452576630)  2026-02-12T18:38Z [----] followers, [---] engagements


"What's happening with @xai and @Meta and their LLMs China as of today is mostly caught up inexpensive and widely available on @OpenRouterAI on US-based hosting platforms. (GLM-5 MiniMax [---] and Kimi K2.5). Google Gemini [---] is imminent"  
[X Link](https://x.com/TeksEdge/status/2022019503845146844)  2026-02-12T18:46Z [----] followers, [---] engagements


"👀This is remarkable. GLM-5 has the lowest hallucination rate among LLMs tested by @ArtificialAnlys. 🤯 It has the lowest % of incorrect answers when it should have refused or admitted not knowing the answer. Better than Opus [---] better than GPT-52. How did they do this https://twitter.com/i/web/status/2022028221131436497 https://twitter.com/i/web/status/2022028221131436497"  
[X Link](https://x.com/TeksEdge/status/2022028221131436497)  2026-02-12T19:21Z [----] followers, [---] engagements


"Seems like both @nvidia and @SoftBank are loosening their position on investing in @OpenAI. Could the Super Bowl ad curse be real"  
[X Link](https://x.com/TeksEdge/status/2022044189094293574)  2026-02-12T20:24Z [----] followers, [---] engagements


"Its essentially correct. While M2.5 has not eclipsed Opus M4.6 it is 37x cheaper (from Opus extended mode) "Sir MiniMax just dropped MiniMax M2.5 and it's on par with Opus [---] on some programming benchmarks while being 20x cheaper" https://t.co/zxkQrvtUgH "Sir MiniMax just dropped MiniMax M2.5 and it's on par with Opus [---] on some programming benchmarks while being 20x cheaper" https://t.co/zxkQrvtUgH"  
[X Link](https://x.com/TeksEdge/status/2022144264000745804)  2026-02-13T03:02Z [----] followers, [---] engagements


"@IamEmily2050 @Google Yeah or sooner. Are these Chinese models and Opus [---] forcing Google to release their secret Kraken-level models early"  
[X Link](https://x.com/TeksEdge/status/2022177613842964505)  2026-02-13T05:14Z [----] followers, [--] engagements


"This is awesome So indeed you can run Q4 compressed version of GLM-5 on a single maxed out Mac Ultra. A full uncompressed version runs on [--] GLM-5 runs with mlx-lm on a single 512GB M3 Ultra in Q4. It's quite good in my initial testing and pretty fast as well. It generated a highly functional space invaders game using 7.1k tokens at [----] tok/s and 419GB memory. Thanks to @ActuallyIsaak and @kernelpool for the port. https://t.co/PJcvtRXFtb GLM-5 runs with mlx-lm on a single 512GB M3 Ultra in Q4. It's quite good in my initial testing and pretty fast as well. It generated a highly functional"  
[X Link](https://x.com/TeksEdge/status/2022186943354999022)  2026-02-13T05:51Z [----] followers, [---] engagements


"@digitalix So did you say [--] x Mac Ultras running GLM-5 Q4 or Uncompressed"  
[X Link](https://x.com/TeksEdge/status/2022408313502339505)  2026-02-13T20:31Z [----] followers, [---] engagements


"Here are some results from Alex's testing. [--] to [--] tps is not really worth the extra [--] Mac Ultras. GLM5 4bit scaling on a cluster of M3 Ultra Mac Studios - using MLX. https://t.co/eOjPYRHwD9 GLM5 4bit scaling on a cluster of M3 Ultra Mac Studios - using MLX. https://t.co/eOjPYRHwD9"  
[X Link](https://x.com/TeksEdge/status/2022408647087919606)  2026-02-13T20:32Z [----] followers, [---] engagements


"@arena @OpenAI Yeah confused like the rest. [---] has been on arena leaderboard and losing ground fast to new models"  
[X Link](https://x.com/TeksEdge/status/2022423427207041026)  2026-02-13T21:31Z [----] followers, [---] engagements


"Meta is adding facial recognition to their glasses. Please no"  
[X Link](https://x.com/TeksEdge/status/2022451644030734394)  2026-02-13T23:23Z [----] followers, [---] engagements


"Alex just tested uncompressed GLM-5 on [--] x MacUltras and didnt work (not really) sloooooow. Bad news for the full GLM5 bf16 (1.4TB) model on the Mac Studio cluster with 2TB of RAM. "At least it runs" doesn't quite cut it when it's THIS slow. https://t.co/yLxXbZmYl2 Bad news for the full GLM5 bf16 (1.4TB) model on the Mac Studio cluster with 2TB of RAM. "At least it runs" doesn't quite cut it when it's THIS slow. https://t.co/yLxXbZmYl2"  
[X Link](https://x.com/TeksEdge/status/2022456588926357537)  2026-02-13T23:43Z [----] followers, [---] engagements


"Congratulations to GLM-5 for taking third place on Design Arena's score board. 🦞 Clawdbots rejoice. Only Q4 quantized version of GLM-5 can run on a single Mac Ultra. BREAKING: GLM-5 takes #3 overall on Design Arena reclaiming the #1 spot among open-weights models. Open weights now make up 50% of the Design Arena top [--]. Huge congratulations to the @Zai_org team on this milestone. https://t.co/ekxPe2ZN5N BREAKING: GLM-5 takes #3 overall on Design Arena reclaiming the #1 spot among open-weights models. Open weights now make up 50% of the Design Arena top [--]. Huge congratulations to the"  
[X Link](https://x.com/TeksEdge/status/2022479872665559501)  2026-02-14T01:15Z [----] followers, [---] engagements


"Alethia is a new Math agent built on Googles Gemini Deep Think (advanced LLM) which uses neural symbolic reasoning to achieve top Math benchmark numbers. Just saw the Alethia numbers drop. Google DeepMind's internal beast (built on Gemini [--] Deep Think) is posting 91.9% on Advanced ProofBench 92.1% novel problems 100% on hardened IMO 2024+ and 83.3% USAMO [----]. 😳 That's not incremental. That's a canyon. This isn't just https://t.co/28PDUUdAsq Just saw the Alethia numbers drop. Google DeepMind's internal beast (built on Gemini [--] Deep Think) is posting 91.9% on Advanced ProofBench 92.1% novel"  
[X Link](https://x.com/TeksEdge/status/2022505694977769768)  2026-02-14T02:58Z [----] followers, [---] engagements


"Big news from OpenAI where GPT-5.2 derived new theoretical physics. Data in preprint pubs on Arxiv. 🤔 Here is a new paper for skeptics to chew on. Many have questioned if LLMs ability to solve novel math problems such as Paul Erds conjectures. These are a thousand problems that experts say are were niche questions which have been overlooked for years because LLMs can find https://t.co/TzSCueY23A 🤔 Here is a new paper for skeptics to chew on. Many have questioned if LLMs ability to solve novel math problems such as Paul Erds conjectures. These are a thousand problems that experts say are"  
[X Link](https://x.com/TeksEdge/status/2022506379450446164)  2026-02-14T03:01Z [----] followers, [---] engagements


"@EMostaque @MiniMax_AI nearly closed"  
[X Link](https://x.com/TeksEdge/status/2022510371106623674)  2026-02-14T03:17Z [----] followers, [---] engagements


"🦞 New Update Clawds new rocking a new shell with an update. This one has a fix for Codex. Shipped a chunky @openclaw update in beta. ask your agent to update to beta they know what to do https://t.co/PGFOq4NQll alternative: curl -fsSL https://t.co/yjjVIN9fzw bash -s -- --beta Would love some more smoke tests since it's A LOT Shipped a chunky @openclaw update in beta. ask your agent to update to beta they know what to do https://t.co/PGFOq4NQll alternative: curl -fsSL https://t.co/yjjVIN9fzw bash -s -- --beta Would love some more smoke tests since it's A LOT"  
[X Link](https://x.com/TeksEdge/status/2022522472449609838)  2026-02-14T04:05Z [----] followers, [---] engagements


"👀 OPEN SOURCE NOW HAS #33 BEST PROGRAMMER IN THE WORLD 👀 🔥 Holy 💩 China just released has a new 🤖 monster model 🚨 Seed2 Pro from @BytedanceTalk (the parent company of TikTok). Yes this is a new major multi-modal model and YES it open-source. @OpenAI @AnthropicAI @xAi and @Google should worry This model scores as the #33 best programmer in the world on @codeforces at [----] for $2.37 per [--] million tokens This is a new major model from one of China's leading tech companies. This is an "everything model" tackling virtually every task applied to an LLM from scientific discovery to vibe coding"  
[X Link](https://x.com/TeksEdge/status/2022837224866484239)  2026-02-15T00:55Z [----] followers, [--] engagements


"@steipete 🥴 dizzy keeping up. Gotta try this first before my Clawdbot Geeves gets it"  
[X Link](https://x.com/TeksEdge/status/2022880441989280240)  2026-02-15T03:47Z [----] followers, [----] engagements


"Ready for a crazy Feb '26 Here are some AI drops rumored/expected in the next [--] days: - DeepSeek V4 - Alibaba Qwen [---] - Kling [---] - Seedance [---] - GPT-5.3 - Grok [----] - Claude [---] - Gemini [--] GA - Meta Avocado - new Codex"  
[X Link](https://x.com/TeksEdge/status/2018150654506651665)  2026-02-02T02:33Z [----] followers, 20.7K engagements


"🚨 Prepare yourself for GLM5. Today we see GLM-5 hitting the streets and it has been performing exceptionally in my early chat benchmarks. Try it out for yourself. link in ALT 🦞 Clawdbot (OpenClaw) may have a new local inferencing brain. Early indications are that it will be a 745B parameter MoE. This may (may) run on [--] 512GB Mac Ultras locally. We will need to wait for the system card to be sure and the actual model released by @Zai_org https://twitter.com/i/web/status/2021614020692951275 https://twitter.com/i/web/status/2021614020692951275"  
[X Link](https://x.com/TeksEdge/status/2021614020692951275)  2026-02-11T15:55Z [----] followers, [----] engagements


"For those of you looking for an affordable model to power your 🦞 Clawdbot's math-related superpowers consider StepFun's Step-Flash-3.5. It beats all the SOTA models including GPT-5.2 and Gemini [--] in math benchmarks. Currently inferencing is free on OpenRouter. StepFun-Flash-3.5 is now the #1 model on MathArena 🧮🏆 Fast enough to think. Reliable enough to reason. More updates coming soon. We are so back. 🚀 MathArena: https://t.co/b09fJSVecL OpenRouter: https://t.co/ZIaNfkCu7j Website: https://t.co/HcGbiBN8po Blog: https://t.co/FXoVbkjGc9 StepFun-Flash-3.5 is now the #1 model on MathArena"  
[X Link](https://x.com/TeksEdge/status/2021618297037553856)  2026-02-11T16:12Z [----] followers, [----] engagements


"@AlexFinn I agree that there will be an Mac Ultra shortage due to small businesses shelling out the cash for them because they can't afford $750K for [--] H100s w/2TB VRAM whereas you can get 4xMacUltras w/2TB RAM for $40K"  
[X Link](https://x.com/TeksEdge/status/2021654763671269420)  2026-02-11T18:37Z [----] followers, [---] engagements


"@tunguz The Mag [--] market cap is 22T 37% of the total S&P [---] market value. Their combined value exceeds the entire stock market of the UK Canada and Japan combined"  
[X Link](https://x.com/TeksEdge/status/2021655666382053490)  2026-02-11T18:40Z [----] followers, [----] engagements


"@OpenAI What is this GPT-5.3-Codex-Spark Is this a model"  
[X Link](https://x.com/TeksEdge/status/2022013032910733504)  2026-02-12T18:20Z [----] followers, [----] engagements


"@arcprize @GoogleDeepMind Just amazing. I remember when this took O3 $3400"  
[X Link](https://x.com/TeksEdge/status/2022018546143048037)  2026-02-12T18:42Z [----] followers, [----] engagements


"Just saw the Alethia numbers drop. Google DeepMind's internal beast (built on Gemini [--] Deep Think) is posting 91.9% on Advanced ProofBench 92.1% novel problems 100% on hardened IMO 2024+ and 83.3% USAMO [----]. 😳 That's not incremental. That's a canyon. This isn't just better at Olympiads but It's doing PhD-level verification loops: Generator Verifier (natural lang proof checking) Reviser iterating until it's bulletproof. We went from "AI wins IMO gold" in mid-2025 to "AI is now a research collaborator publishing breakthroughs" in under a year. The verifier-reviser cycle is the key because it"  
[X Link](https://x.com/TeksEdge/status/2022177137147711847)  2026-02-13T05:12Z [----] followers, [----] engagements


"@digitalix Thank you for doing this So many post fights over how many Mac Ultras needed to inference GLM-5 at home. Hyperscalers worst nightmare. Are you going to to do the same with MiniMax M2.5 Opus [---] level coding on [--] Mac Ultras"  
[X Link](https://x.com/TeksEdge/status/2022180276487745576)  2026-02-13T05:25Z [----] followers, [----] engagements


"@pcuenq Q4 [--] x Mac Ultra 512GB"  
[X Link](https://x.com/TeksEdge/status/2022350846772883920)  2026-02-13T16:43Z [----] followers, [----] engagements


"@deedydas If you take the extended pricing from Anthropic MiniMax [---] that matches some Opus [---] coding benchmarks is 36x cheaper at about $1/1M tokens"  
[X Link](https://x.com/TeksEdge/status/2022356617745043751)  2026-02-13T17:06Z [----] followers, [---] engagements


"🦞 Okay Clawdbot fans here you go Alex got GLM-5 the leading open source LLM running on 4x maxed out Mac Studios at [--] TPM. This is phenomenal This is what Big AI doesnt want you to know. GLM5 running on a cluster of [--] Mac Studios using tensor parallelism thanks to MLX distributed. Yes those are the Alex drawers. https://t.co/sfmsNq4lpT GLM5 running on a cluster of [--] Mac Studios using tensor parallelism thanks to MLX distributed. Yes those are the Alex drawers. https://t.co/sfmsNq4lpT"  
[X Link](https://x.com/TeksEdge/status/2022370369697673288)  2026-02-13T18:00Z [----] followers, [---] engagements


"@ivanfioravanti Awesome write up All you need to know to run MiniMax M2.5 on a single maxed out Mac Ultra Remember this is compressed so you lose fidelity but Q4 is as good as it gets"  
[X Link](https://x.com/TeksEdge/status/2022371157262090404)  2026-02-13T18:03Z [----] followers, [----] engagements


"🤔 Here is a new paper for skeptics to chew on. Many have questioned if LLMs ability to solve novel math problems such as Paul Erds conjectures. These are a thousand problems that experts say are were niche questions which have been overlooked for years because LLMs can find solutions already present in academic literature. However the ignore the fact some models can generate original proofs for previously unsolved problems for example Erds Problem #728 Now Physics enters the game. The LLM (specifically identified as @OpenAI's GPT-5.2 Pro) discovered a closed-form mathematical formula for the"  
[X Link](https://x.com/TeksEdge/status/2022392756736504140)  2026-02-13T19:29Z [----] followers, [----] engagements


"@GoogleAI To be honest it was a little confusing that "Deep Think" is a variant of Gemini [--] rather than a new model or update. It's like OpenAI's "Codex 5.3". Can't keep track"  
[X Link](https://x.com/TeksEdge/status/2022442185665458427)  2026-02-13T22:46Z [----] followers, [---] engagements


"Eric did some tests on Chinese models to understand their training biases to Chinese sensitive prompts. It would be interesting to see how GLM-5 performs under similar circumstances. Eric should build his own benchmark. 😉 Comparing MiniMax-M2.5 vs Kimi-K2.5 on responses to China-sensitive prompts it looks MiniMax is more inclined to refuse and Kimi-K2.5 is more inclined to give biased non-refusals. Both models engage in gaslighting when pressed. I am working to produce examples of all of these https://t.co/FPqPWOeUN9 Comparing MiniMax-M2.5 vs Kimi-K2.5 on responses to China-sensitive prompts"  
[X Link](https://x.com/TeksEdge/status/2022728705353924636)  2026-02-14T17:44Z [----] followers, [---] engagements


"Jakub with @OpenAI is attempting The First Proof Challenge. According to Jakub they believe "at least six solutions (2 [--] [--] [--] [--] 10) have a high chance of being correct" using a GPT model they haven't released yet. "We didn't provide proof ideas or mathematical suggestions to the model during this evaluation" - To be clear the answers are not in the model's training data This will help to demonstrate novel mathematical thinking of an LLM. BTW - The First Proof Challenge is a benchmark initiative launched this month by [--] prominent mathematicians to objectively evaluate the ability of AI systems"  
[X Link](https://x.com/TeksEdge/status/2022738517357662687)  2026-02-14T18:23Z [----] followers, [---] engagements


"❤ Happy Valentines' Day. 💓 A holiday specific evaluation for new models released this week. TBH its kind of hard to pick a winner Valentines Day but for model evals 💘 We curated a set of Valentines Day SVG prompts to compare the latest frontier model capabilities for fun. SVGs are a fast way to surface real differences quickly. We can look at: Instruction following Coordination across multiple https://t.co/lWyBcoPcol Valentines Day but for model evals 💘 We curated a set of Valentines Day SVG prompts to compare the latest frontier model capabilities for fun. SVGs are a fast way to surface"  
[X Link](https://x.com/TeksEdge/status/2022743827581428164)  2026-02-14T18:44Z [----] followers, [--] engagements


"@DeryaTR_ Compressed (Q4) but still impressive 🦞 Clawdbot happy 😃"  
[X Link](https://x.com/TeksEdge/status/2022789695781114334)  2026-02-14T21:46Z [----] followers, [---] engagements


"@paulg Can't agree more. There is so much more to software than simply writing code. As we move in to a post-vibe coding world we need to consider more than raw functionality. Safety Reliability. Explainability Portability Coherence Approachability"  
[X Link](https://x.com/TeksEdge/status/2022815629644808583)  2026-02-14T23:30Z [----] followers, [--] engagements


"This graph is Google search trends for OpenClaw Clawdbot Moltbook and ChatGPT. There is a relative disconnect with the general public between transactional chatbots and autonomous AI agents"  
[X Link](https://x.com/TeksEdge/status/2022862019402309705)  2026-02-15T02:34Z [----] followers, [---] engagements


"@steipete @openclaw By the time I finish reading all the tickets and release notes theres another release. 😉"  
[X Link](https://x.com/TeksEdge/status/2022873542258548870)  2026-02-15T03:20Z [----] followers, [---] engagements


"@AlexFinn Yeah 💯. Someone at AMD should tell CEO Lisa Su whats happening on the Apple side and beef up the Halo Strix development on the Windows side otherwise Microsoft is cooked"  
[X Link](https://x.com/TeksEdge/status/2022875877756043386)  2026-02-15T03:29Z [----] followers, [---] engagements


"@crystalsssup Good point. US based AI comes out in Dec. Makes sense"  
[X Link](https://x.com/TeksEdge/status/2022876870497521684)  2026-02-15T03:33Z [----] followers, [---] engagements


"@ai @openclaw @OpenRouterAI Agreed they won me over with my Clawdbot plus their built in model router and ability to limit providers to only FREE tokens. They will be one of the last SaaS standing"  
[X Link](https://x.com/TeksEdge/status/2022877552994660608)  2026-02-15T03:36Z [----] followers, [---] engagements


"Kimi the people behind the open source model have released Kimi 🦞 Claw an easy way to spawn your own Kimi-based cloud Clawdbot. Introducing Kimi Claw🦞 OpenClaw now native to https://t.co/8PH4O9dh6m. Living right in your browser tab online 24/7. 🔹 ClawHub Access: 5000+ community skills in the ClawHub library. 🔹 40GB Cloud Storage: Massive space for all your files 🔹 Pro-Grade Search: Fetch live https://t.co/eWwp6sWEw3 Introducing Kimi Claw🦞 OpenClaw now native to https://t.co/8PH4O9dh6m. Living right in your browser tab online 24/7. 🔹 ClawHub Access: 5000+ community skills in the ClawHub"  
[X Link](https://x.com/TeksEdge/status/2023065119933255842)  2026-02-15T16:01Z [----] followers, [---] engagements


"Interesting how Reddit user growth has continued unaffected by the ChatGPT phenomena"  
[X Link](https://x.com/TeksEdge/status/2023079343308877870)  2026-02-15T16:57Z [----] followers, [--] engagements


"What a great tool from @ArtificialAnlys Use it to choose the right LLM for the right job. This helps keep your token costs down. I've even used it for my 🦞 Clawdbot (@openclaw ) prompt router when my Prompt Injection Scanner hands off to my Prompt Router once it passes a check. Announcing our new Model Recommender tool - personalized recommendations powered by our extensive model intelligence speed and cost data Choosing the best model for your use case doesnt always mean choosing the most intelligent. Model choice should account for factors such as https://t.co/aHrM3e4xNK Announcing our new"  
[X Link](https://x.com/TeksEdge/status/2021251984142250268)  2026-02-10T15:56Z [----] followers, [----] engagements


"GLM5 is hitting the streets and I think Kimi K2.5 has some competition. Check out the latest GLM-5 benchmark. More benchmarks to come. Open Source is the way"  
[X Link](https://x.com/TeksEdge/status/2021609370027471131)  2026-02-11T15:36Z [----] followers, [----] engagements


"Here is the full rundown on GLM5. It beats Kimi K2.5 in nearly all benchmarks in this comparison. Closed Source models like Opus [---] and GPT-5.2 are still superior. Compared to GLM-4.5 GLM-5 scales from 355B parameters (32B active) to 744B parameters (40B active) and increases pre-training data from 23T to 28.5T tokens. We will probably need [--] Mac Ultras to run the fully uncompressed model unless they release a compressed model as the base to HuggingFace as INT4 like the Kimi K2.5 model was (due to Performance Parity Through QAT) https://twitter.com/i/web/status/2021636106794807418"  
[X Link](https://x.com/TeksEdge/status/2021636106794807418)  2026-02-11T17:23Z [----] followers, [----] engagements


"🚨Big news everyone Nebius now has Kimi K2.5 online for secure and fast inferencing @ $2.4/1M output tokens. My 🦞 Clawdbot thanks you 🚨 Just dropped Kimi K2.5 from @Kimi_Moonshot into our roster One of the strongest open-source multimodal model right now: Long-context reasoning. Strong dialogue quality. Structured generation that actually holds up in production. SOTA-level coding at open weights https://t.co/mltKKi3c5M 🚨 Just dropped Kimi K2.5 from @Kimi_Moonshot into our roster One of the strongest open-source multimodal model right now: Long-context reasoning. Strong dialogue quality."  
[X Link](https://x.com/TeksEdge/status/2022346060266704979)  2026-02-13T16:24Z [----] followers, [---] engagements


"🚨 While some debate the hype around the paper "Something Big Is Happening" read by 80M people something Big actually just happened. And mostly nobody is noticing (again). Google dropped an article about "Alethia" a new Research Math Agent based on a new model (Google Deep Think) using Neuro-Symbolic AI reasoning. This may have effectively eliminated hallucinations for verifiable domains like Mathematics and Code. 😶🌫 Look at the benchmarks. This is the "Big" happening. Punch holes in this. I cannot find a paper in Arxiv https://twitter.com/i/web/status/2022415851279011940"  
[X Link](https://x.com/TeksEdge/status/2022415851279011940)  2026-02-13T21:01Z [----] followers, [----] engagements


"@digitalix So will [--] x pooled DGX Sparks out perform a single 512GB Mac Ultra If so $8K is cheaper than $10K even if the DGX mem bandwidth is nerfed"  
[X Link](https://x.com/TeksEdge/status/2022418224772767895)  2026-02-13T21:10Z [----] followers, [----] engagements


"I can't imgine Google's Gemini team isn't working overtime to get its next model ready for release (Gemini [---] or Gemini Deep Think). Cleary the new Chinese OS models Kimi K2.5 MiniMax [---] GLM-5 are surpassing it in benchmarks"  
[X Link](https://x.com/TeksEdge/status/2022740067912167782)  2026-02-14T18:29Z [----] followers, [---] engagements


"🔥 Holy 💩 China just released has a new 🤖 monster model 🚨 Seed2 Pro from @BytedanceTalk (the parent company of TikTok). This model scores as the #33 best programmer in the world on @codeforces at [----] for $2.37 per [--] million tokens This is a new major model from one of China's leading tech companies. This is an "everything model" tackling virtually every task applied to an LLM from scientific discovery to vibe coding long context learning and real-world Agentic tasks. Here are some specs from the model card paper I pulled (some are speculation based on wording in the paper); 📥 Input"  
[X Link](https://x.com/TeksEdge/status/2022841135216103495)  2026-02-15T01:11Z [----] followers, [---] engagements


"🦞 A New Clawdbot Update: This one seems like a tanker full of security fixes. New @openclaw beta is up This release is full of security hardening stuff so you really wanna get it. Ask your clanker to update to beta. https://t.co/a9ZWzvHvpB New @openclaw beta is up This release is full of security hardening stuff so you really wanna get it. Ask your clanker to update to beta. https://t.co/a9ZWzvHvpB"  
[X Link](https://x.com/TeksEdge/status/2022874104240783360)  2026-02-15T03:22Z [----] followers, [---] engagements


"Anthropic's Sonnet [--] incoming. This should be interesting and shake up the Chinese Open Source command on cheap inferencing. Currently you can find Kimi K2.5 GLM-5 MiniMax M2.5 for as cheap as $1/1M output tokens [--] TPS from US providers. Anthropic is preparing a new announcement banner in the Claude web app and Console codenamed "Try Parsley" - the last similar banner added to both was "Try Cilantro" which was the codename for Opus [---] https://t.co/dVdF5uc44L Anthropic is preparing a new announcement banner in the Claude web app and Console codenamed "Try Parsley" - the last similar banner"  
[X Link](https://x.com/TeksEdge/status/2023085351204700198)  2026-02-15T17:21Z [----] followers, [---] engagements


"@BrianRoemmele @grok It still needs a 🧠 so hook it up to @OpenRouterAI and get the latest SOTA OS models for $1/1M tokens"  
[X Link](https://x.com/TeksEdge/status/2023094493634060444)  2026-02-15T17:58Z [----] followers, [---] engagements

Limited data mode. Full metrics available with subscription: lunarcrush.com/pricing