[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]

@Hangsiin Avatar @Hangsiin NomoreID

NomoreID posts on X about token, automatic, generated, ai the most. They currently have XXXXX followers and XXX posts still getting attention that total XXXXX engagements in the last XX hours.

Engagements: XXXXX #

Engagements Line Chart

Mentions: XX #

Mentions Line Chart

Followers: XXXXX #

Followers Line Chart

CreatorRank: XXXXXXXXX #

CreatorRank Line Chart

Social Influence

Social category influence technology brands finance

Social topic influence token, automatic, generated, ai, banana, nano banana, opera, applied

Top accounts mentioned or mentioned by @openai @sama @andrewcurran @giffmana @btibor91 @presidentlin @windsurfai @morqon @xiaohuazhai @kolesnikov @hcsolakoglu @abacaj @tszzl @fouadmatin @bradlightcap @googledeepmind @mjdramstead @versesai @therealadamg @gdb

Top Social Posts

Top posts by engagements in the last XX hours

"I am currently testing GPT 5.2s long-context capabilities in my own way by giving it the full script of a specific game (Expedition 33) and asking for a detailed analysis. So far the results have been quite impressive. Compared with GPT X it feels like a very noticeable improvement: it picks up subtle details that are easy to miss offers interpretations you would normally expect only from someone who has played the game very deeply and shows almost no hallucinations. (GPT X in contrast did get pulled into a few hallucinations.) I even had a slightly uncanny moment just now when it did not"
X Link 2025-12-11T21:56Z 2552 followers, 108.5K engagements

"GPT-5.2 (high) ranked first in my Korean Sator Square Test (KSST) beating Gemini XXX Pro by a meaningful margin. (165.9/190 vs 155.3/190) At first I did not plan to test it because the API cost felt burdensome. However after running one sample with GPT-5.2 in ChatGPT and being impressed by the result I became curious and decided to proceed with the test. For AVG TOKEN GPT-5.2 API occasionally times out. Since the generated output remains in the ChatGPT log I manually copied and pasted only the results. Because of that the AVG TOKEN values became inaccurate so I edited them manually this time."
X Link 2025-12-13T02:34Z 2552 followers, 23.1K engagements

"@inductionheads gpt-3.5-turbo claude-3.5-haiku"
X Link 2025-12-05T17:07Z 2545 followers, XXX engagements

"@inductionheads sorry it was claude-3-haiku"
X Link 2025-12-05T17:09Z 2545 followers, XXX engagements

"@HystoEU I have not been able to use GPT XXX in the ChatGPT app yet so I tested it through the Codex CLI in an empty directory. Because of that it is a bit difficult for me to share the conversation (and it would also contain spoilers.) Sorry about that"
X Link 2025-12-11T23:06Z 2545 followers, 1361 engagements

"@tszzl I compiled all in-game dialogue into a single .txt file and used it as context. It contains only the full dialogue with no subjective interpretations or explanations of the setting"
X Link 2025-12-11T23:44Z 2553 followers, 3479 engagements

"@tszzl For reference this wasnt a one-off. I ran the same test using the exact same approach on earlier models that marketed long context as a key strength as well and GPT-5.2 stood out among them"
X Link 2025-12-11T23:47Z 2547 followers, 1674 engagements

"For reference I recently learned that using the Codex CLI can be a good approach for analyzing long-context inputs. It supports up to 400000 tokens of context and with automatic context compression it can handle even larger analyses. Also the Plus subscription alone provides sufficient usage limits which makes it a cost-effective option as well"
X Link 2025-12-12T00:35Z 2547 followers, 2665 engagements

"@AcerFur When the timeout occurred how many output tokens(reasoning) had been generated"
X Link 2025-12-13T17:06Z 2550 followers, XXX engagements

"When asked to speak without exaggeration about AI advancements over the next XX months Demis Hassabis(@demishassabis) highlighted the following three points: X. A lot of progress in multimodality: We are currently gaining very interesting synergy effects as models become multimodal. A recent example of this is "Nano Banana Pro." In particular as video and language models combine next year we will see a very interesting combination of capabilities. X. Advancement of 'World Models': The development of models such as Genie-3. X. Reliable Agents: While the industry has discussed agents"
X Link 2025-12-05T20:57Z 2554 followers, 336.2K engagements

"Hello Thank you for your interest. The underlying idea of my test is very simple. You may or may not be familiar with the Sator Square but in essence it is an arrangement of letters in an n n grid such that the text reads the same horizontally and vertically. ex) Seed: 'SATOR' SATOR AREPO TENET OPERA ROTAS My test goes beyond merely requiring the horizontal and vertical readings to match. It asks an LLM to write a special type of poem that also satisfies those structural constraints. Because the conditions are extremely demanding it is easy to adjust the difficulty by making the given lines"
X Link 2025-12-13T11:33Z 2551 followers, XXX engagements

"Im looking forward to the Gemini XXX Flash and Pro-based audio models. Just as nanobanana did for image generation I hope they can deliver a step-change leap forward in audio generation"
X Link 2025-12-13T13:25Z 2554 followers, 1483 engagements

"I largely agree with what you said. In general it makes sense for simpler queries to use a smaller reasoning budget. The issue is that with the current model when almost no reasoning is used for those simple problems the final response quality can still drop noticeably. As a result for queries where some reasoning used to be applied GPT-5.2 can end up giving weaker answers than earlier models which ultimately makes the overall experience worse"
X Link 2025-12-14T10:25Z 2554 followers, XXX engagements