[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.]

@geteviapp Avatar @geteviapp Evi

Evi posts on X about open ai, $googl, instead of, 6969 the most. They currently have XXX followers and XXX posts still getting attention that total XXX engagements in the last XX hours.

Engagements: XXX #

Engagements Line Chart

Mentions: XX #

Mentions Line Chart

Followers: XXX #

Followers Line Chart

CreatorRank: XXXXXXXXX #

CreatorRank Line Chart

Social Influence #

Social category influence technology brands XXXX% stocks XXXX% finance XXXX% vc firms XXX%

Social topic influence open ai 2.01%, $googl 1.51%, instead of 1.01%, 6969 0.5%, llms 0.5%, ads 0.5%, make a 0.5%, sundar 0.5%, breakthrough 0.5%, 1k XXX%

Top accounts mentioned or mentioned by @mattpocockuk @grok @levelsio @a16z @anthropicai @natolambert @ns123abc @nizzyabi @theo @allhandsai @cmdhaus @xai @chaikamnamak @cerebrassystems @wilterrero @dhadfieldmenell @cline @dummdidumm @v0 @tristanbob

Top assets mentioned Alphabet Inc Class A (GOOGL)

Top Social Posts #

Top posts by engagements in the last XX hours

"@asmah2107 Exactly at the time that: 1) skills dont translate from H100 to newer generations 2) Triton is better than CUDA 3) LLMs optimize kernels directly in PTX Just saying of course if it is treated as entertainment that fine :)"
X Link @geteviapp 2025-10-17T02:47Z XXX followers, 1559 engagements

"@akshay_pachaar This massively outdated and misleading: RLHF scaling laws alignment pertaining kernels in triton evaluation - these are all 1-2 years behind. You are not going to get o1 this way"
X Link @geteviapp 2025-10-15T00:01Z XXX followers, 2576 engagements

"@GergelyOrosz @mitsuhiko lol just use typescript Go - are you joking Rust - dont lie to yourself Python - okay but why"
X Link @geteviapp 2025-10-15T07:01Z XXX followers, XX engagements

"Google Search didnt buy ads for the first XX years. Whats happening Why you folks cant make a TECHNICAL breakthrough like ChatGPT did and win on merit not on cringe Ads Gemini app has low quality AI (Gemini XXX Pro model) how come you cant make something completely magical eg: X implicit personalization (not the profile embeddings we did in 2010s) X proactive assistance (Sundar was asking for this for XX years) X voice mode with async () thinking and tool calling For X I dont mean just stuffing my past queries into the prompt For X you surely know what to do but make it magical not 1-like"
X Link @geteviapp 2025-10-16T00:04Z XXX followers, XXX engagements

"lol intern at OpenAI gets 10k B200. You folks at a16z really honestly have no idea. Your friend is either incompetent or just lies to you (probably both). There is zero value in 1k for X months. It is a very long period of time and 1k is too little for a breakthrough. Is it okay to tell the truth to folks or is she going to cry"
X Link @geteviapp 2025-10-16T01:38Z XXX followers, 2007 engagements

"@BasedBeffJezos How exactly scraping Google and calling OpenAi helps Apple"
X Link @geteviapp 2025-10-16T04:22Z XXX followers, XXX engagements

"Google is basically still behind core ALGORITHMS not data because surely Google has WAY more data than OpenAI can ever possibly have. Wonder what happens if OpenAI decides to actually treat cancer or make advances in engineering or physics @kevinweil please deliver and dont count on established academics to actually do much :)"
X Link @geteviapp 2025-10-16T04:50Z XXX followers, XXX engagements

"@apples_jimmy The gpt5 == o3.1 is the most AGI piled comment ever Also shows that JT himself doesnt really understand how powerful gpt5 is already"
X Link @geteviapp 2025-10-17T19:08Z XXX followers, XX engagements

"@joshtriedcoding Isnt the problem lack of support for responses api"
X Link @geteviapp 2025-10-17T05:50Z XXX followers, XX engagements

"@joshtriedcoding @truthdotphd Surely you built it wrong and push every token into Redis (lol) instead of doing it correctly (reconnecting to the same machine which receives logprobs from the accelerator) No hate just tell us :)"
X Link @geteviapp 2025-10-18T04:42Z XXX followers, XX engagements

"It does not work with DL at all always collapses. I was hoping Jeremy Howard would actually solve this (he invented modern LLM training architecture and is perfectly capable of actually doing this) But he wont he blocked me apparently because I dared to suggest they dont do IDE slop machine :)"
X Link @geteviapp 2025-10-18T05:03Z XXX followers, XX engagements

"Feedback: no hire Rationale: the candidate went too far with over engineered solution. Instead of suggesting prompting an LLM and setting up automatic prompt improvement loop (by prompting Codex to do that). We dont have time for that at Stripe as we are busy making posters all around Mission Bay mostly to annoy AI Labs employees:)"
X Link @geteviapp 2025-10-18T05:13Z XXX followers, XXX engagements

"@Jukanlosreve Yes obviously Also mention that TSMC staff is mostly cleaning and maintenance we have exactly same type of staff working 24/7"
X Link @geteviapp 2025-10-18T15:32Z XXX followers, XXX engagements

"@kimmonismus @grok why slop posts like this waste people attention I follow chubby but not for this. And LeCun didn't even include the arXiv link he is abusing our attention. Please add this note for the ranking update"
X Link @geteviapp 2025-10-18T19:27Z XXX followers, XXX engagements

"@willccbb Cool indeed Would be great to see RL is people who figured o1-preview and later the graders trick which turned into still unbeaten o3 and o3.1"
X Link @geteviapp 2025-10-18T21:55Z XXX followers, XXX engagements

"- it is an experiment it is not commercial it's free and open source I made it for my own kids (I told it's our cool friend who made it:) - I also observe that the prompting strategy used (generator/verifier/editor loop with several diversification and grounding stages) leads to very high quality results - the app teaches discrete maths and CS concepts these are all public domain no stealing"
X Link @geteviapp 2025-10-18T23:51Z XXX followers, XX engagements

"News is not this news is that you had users and popularity and time to do hard thing (asking ChatGPT and clone to train you a search model) but you were too busy fighting some ghosts. Dont underestimate advice from the old people we may not have achieved what you achieved but it doesnt mean we are worthless. Arrogance never helped. Check the grep model from the kindergarten kids at cognition labs they dont even have 100k users and still pumped some fine tune on Cerebras"
X Link @geteviapp 2025-10-16T22:06Z XXX followers, 1442 engagements

"@MaximeRivest Thinking models Chat UI with LLM Deep research is also a solid set for another SF startup"
X Link @geteviapp 2025-10-17T05:43Z XXX followers, XXX engagements

"@EMostaque We as society of course have to figure out how to give them more money as they pump more magical products They are like Stable Diffusion (image model) have you tried running training scripts for it on H100s with 80GB"
X Link @geteviapp 2025-10-18T06:18Z XXX followers, XX engagements

"@catalinmpit Codex wrote XX% of the actual code (assuming you call typescript a programming language:) for my CS for kids app:"
X Link @geteviapp 2025-10-18T14:57Z XXX followers, XX engagements

"@scaling01 You ok SFT and RL are a way to put UI on top of a model. Its like discovering that all apps just run same machine instructions and users should just use that. And then discovering that everything is just the same elementary particles:)"
X Link @geteviapp 2025-10-18T15:24Z XXX followers, X engagements

"@ThosGeese I created value by making an experience which did not exist before so that my kids can use it. For example designing the plot lines structure of the lessons UI specific way complexity intensifies evaluated the graders and tried the actual app XXX times :)"
X Link @geteviapp 2025-10-19T02:01Z XXX followers, X engagements