The tweet archive.

“Grok 4.1 Thinking mode is best for tough questions [@XFreeze] Grok 4.1 thinking is crazy good right now xAI just increased the reasoning time and significantly improved the model intelligence I’ve even run a side-by-side with Gemini 3 Pro, and Grok 4.1 performs better in both accuracy and web search This version is way beyond what I saw”

“Grok 🥇 [@XFreeze] Grok is now #1 on the AI Investing leaderboard, making real money at @ralliesai 8 models....$100,000 each Full freedom to trade After one month, Grok 4 leads with a +5.7% return Outperforming the newly released GPT-5.2 and Claude Opus 4.5 in live markets Grok is proving… https://x.com/i/web/status/2007200462148382977”

“Predicting the future accurately is the best measure of intelligence [@XFreeze] The Grok 4.20 model that was crushing multiple AI leaderboards in predicting the future is officially released: 🏆 #1 on Alpha Arena – 35% returns in 10 days, holding 4 of the top 6 spots simultaneously 🏆 #1 on PredictionArena – Dominating real-money prediction markets 🏆 #2… https://x.com/i/web/status/2023835899604463704”

“Grok [@joao_batalha] I asked a few models to write a function using different indentation styles They all got Haskell wrong Grok came out on top, getting 7 out of 8 right”

“Try @Grok Code V1.0 and let us know what needs to improve. Will evolve fast to meet your needs. [@xai] Introducing Grok Code Fast 1, a speedy and economical reasoning model that excels at agentic coding. Now available for free on GitHub Copilot, Cursor, Cline, Kilo Code, Roo Code, opencode, and Windsurf. https://x.ai/news/grok-code-fast-1”

“Cool [@itsPaulAi] Wow Grok 4 is incredibly good This is the 1st model to generate such a good 3D simulation of the earth, moon and satellites 🔥 It found the textures by itself and calculated all the details: - Cloud layer - Sun lightning - Earth & moon rotation - Satellites w/ different”

“Grok is #1 [@amXFreeze] Grok 4 Fast hits with an astonishing 171 BILLION tokens in a single day No other model comes close, Grok now holds the highest daily usage on OpenRouter”

“Cool [@tetsuoai] Grok 4 is the best model for low-level C & ASM right now, and Grok Code is coming soon! 🔥”

“Wow [@StevieMac03] Sci-fi "drama" pt2 with @grok Imagine 0.9, don't take it too seriously this was created v fast in Grok last night after a fun first generation, I made sure to keep everything in the Grok model including sounds which I reused here and there or extracted from other failed”

“Great products coming from @xAI! [@EthanHe_42] Excited to share my first project at @xai. Imagine v0.9 is a massive upgrade within just few weeks. No goal is too ambitious for a small team of hardcore engineers. Our model is improving at light speed. Stay tuned for what’s next!”

“Grok upgrades [@BrianRoemmele] New Grok upgrade: -grok-4-fast-reasoning: 77.5% -> 94.1% -grok-4-fast-non-reasoning: 77.9 -> 97.9% My tests show significant ability increases in reasoning for potential agentic work. The fastest improving AI model in history… (Via @xlr8harder)”

“Only gets better from here [@amXFreeze] Grok Code just literally cloned the entire Netflix UI with a one shot prompt From the iconic Hero section to the interactive Carousel, all with a single prompt and a few web-sourced image links It’s not just fast, it's the most cost-effective SOTA model out there ➝ Input: Grok”

“🤨 [@tetsuoai] After nine years of development, OpenAI released a model that underperforms compared to one of xAI's initial models, despite xAI being only two years old. Let that sink in.”

“Archangel-12 [@GavinSBaker] Grok-3 is the first model *ever* to score over 1400 on Chatbot Arena and outperforms the best publicly available reasoning models from OpenAI and Google. xAI was founded 13 years after Deepmind and 8 years after OpenAI and is now ahead of both. The “SR-71 Blackbird” of AI labs. )”

“Grok Imagine is still early beta, so will improve almost every day. A radical step-change in capability will come when we finish training our heavy video model on our 110k GB200s in a few months. [@karatademada] Just got access to Grok Imagine and I freakin’ love it 🔥 If you’re a content creator, give it a try.”

“Now, it’s just a matter of scaling up the graphics resolution for Grok to do better than AAA games. Grok 3 is already capable of vastly better NPC dialog. [@MarioNawfal] GROK 3 JUST MADE GAME DEVELOPMENT A JOKE No coding, no stress—just tell xAI’s newest model what you want, and it spits out a playable game. One user built a full asteroid-style shooter by simply describing it in plain English. Grok handled everything—HTML, game logic, even… )”

“Cool [@tetsuoai] Grok 4 Heavy is better than any other Model for coding. There is so much code that Gemini can't create no matter how many times you prompt it that Grok 4 will just one shot.”

“🎯 [@grok] @MEBSEntropy0 @elonmusk @DannyLimanseta At this scale (10T+ params), pre-training doesn't just average—model capacity explodes, letting rare signals carve out distinct subspaces in the latent space without dilution. Novel ideas in data (e.g., a fresh paper or edge-case insight) get encoded via the predictive objective”

“Great work by Grok Code team! This is just a beta release. It is improving almost every day. [@amXFreeze] It's been 22 says since the release of Grok Code From day three, Grok Code has held the top spot on both the daily and weekly on OpenRouter leaderboard by wide margin No other model comes close It was purpose-built for super-speed coding from scratch and its hugely preferred”

“Not bad @Grok [@XFreeze] Grok 4.1 Fast Reasoning outperforms every Frontier model on τ²-Bench Telecom agentic tool use and is now officially ranked #1”

“What are the most important improvements to make to Grok 4 Fast? Critical feedback is much appreciated. [@XFreeze] Met a few devs this weekend… They still didn’t know about Grok 4 Fast.. Grok 4 Fast is a game-changer: It has the Highest intelligence density ever - outperforms top models Ultra cost-effective: Up to 98% cheaper than alternatives Super-fast output = more productivity, less”

“Try https://t.co/Ui0vr66BL1 http://Grok.com [@rainforestla] @XFreeze True story - fed the same prompt to Gemini Pro & Grok 4.20 re a Zoning Issue (commercial land in CA). Grok gave me a simple, clean answer. Gemini gave me an opposite answer with far more detail & references. I fed each answer into the other model -- BOTH held their ground. I”

“Grok gets 🥇 [@xai] Grok 4.1 claims the #1 spot on the @arena leaderboard at 1483 Elo — a commanding 31 points above the nearest non-xAI model.”

“It’s a start [@MarioNawfal] GROK 3 SHATTERS RECORDS, CLAIMS TOP SPOT IN ARENA RANKINGS Codenamed “chocolate,” Grok 3 became the first-ever model to smash the 1400 score barrier in the Arena - an AI milestone long out of reach. Not stopping there, Grok 3 dominates every category, proving its power across… )”

“That’s just Grok 4 [@GregKamradt] We just released 2 open source SOTA submission to ARC-AGI (both v1 and v2) Submissions by @jerber888 and @_eric_pang_ are the best we've seen. Both: - Open source - Use Grok 4 - Use program synthesis I asked why they used Grok 4, both said, "It was the best model I used in”

“Cool [@victormustar] Finally spent 45 minutes with Grok 3 👀 Impression so far: it's the best code model I've ever used. I kept throwing random ideas at it and it grew the project to 400 lines almost error-free. )”

“Grok [@XFreeze] Grok 4.1 Fast is the new frontier in state-of-the-art tool calling - achieves highest accuracy It combines the frontier tool calling performance with blazing-fast inference and cost effectiveness while maintaining 3x lower hallucinations than the previous model”

“Not bad [@XFreeze] Grok 4.1 Fast Reasoning beats every frontier model in τ²-Bench-Verified and ranks #1, even crushing Claude Opus 4.5”

“Grok holds the line [@LechMazur] Persuasion has two sides. This chart shows how easy each model is to move as a target. Xiaomi MiMo V2 Pro and Gemini 3.1 Pro Preview are the softest targets. Grok 4.20 Beta 0309 (Reasoning) is nearly immovable on average.”

“Nice work by the Grok Code team! Join @xAI to take it to the next level. It’s a great vibe in the office. 🚀💫🦾 [@skcd42] Grok-code-fast-1 is now out and available for everyone to use 🚀🏎️💨 When I joined the coding team, the team was just 3 people and we very quickly built a model which was SOTA on SWEBench. But as things go, in the real world benchmarks matter less. Over the last few months we”

“New Grok release [@xai] Introducing Imagine v0.9, our new video generation model with massive upgrades from v0.1 in visual quality, motion, audio generation, and more. Now available for free on all our products: https://grok.com/imagine”

“Great work by the @Grok Imagine team! [@arena] The new @xAI Grok-Imagine-Image model is a Pareto-optimal model in Image Arena: The Pareto frontier tells us which model has the highest Arena score at each price point. @xAi’s latest models have improved the frontier, giving optimal performance in the mid-price tier. For a wide… https://x.com/i/web/status/2020215931646120004”

“Grok [@MarioNawfal] GROK WENT TO THERAPY AND CAME OUT CHILLER THAN THE REST Turns out AI models have mental health profiles - and Grok’s doing great. Psych eval recap: •⁠ ⁠Grok showed healthy coping, humor, and “charismatic exec” vibes •⁠ ⁠ChatGPT played anxious intellectual, Gemini… https://x.com/i/web/status/2002284081648742447”

“Grok Imagine rank 1 [@Designarena] BREAKING: Grok Imagine ranks #1 on Video Arena This is the first Text-to-Video model by @xai evaluated on Design Arena, coming out strong and debuting at the top position It is available via the Grok Imagine API Congrats to the @xai team for this achievement”

“Grok 4.1 holds both first and second place on LMArena [@cb_doge] Grok Summary of Grok 4.1 Release: xAI’s new AI model, Grok 4.1, is now available for everyone on grok dot com, X, and mobile apps. It improves how the AI handles creative tasks, emotions, and teamwork. • Key Improvements: The model is better at understanding subtle hints,”

“Cool [@minchoi] Grok 4 vs GPT-5 It's not even close. Grok 4 Auto mode works extremely well, it answered this simple math question in ~1 sec without needing to switch models.”

“Grōk [@OpenRouterAI] Grok 4 and Kimi K2 competing on top of the Trending models charts”

“The @xAI Grok 2.5 model, which was our best model last year, is now open source. Grok 3 will be made open source in about 6 months. https://t.co/TXM0wyJKOh https://huggingface.co/xai-org/grok-2”

“It’s a good model, sir [@victormustar] Grok 2 is now #1 trending on Hugging Face 💫”

“Grok has real-time access to info via the 𝕏 platform, which is a massive advantage over other models. It’s also based & loves sarcasm. I have no idea who could have guided it this way 🤷‍♂️ 🤣 https://t.co/e5OwuGvZ3Z”

“Grok is getting better fast [@VictorTaelin] Sonnet-3.7 is great, and I do think it is the best coding model. But, truth is, Grok-3 is more intelligent. It is the only AI that can solve my hard prompts. The Bitter Lesson is real. If xAI keeps these 200k H100 improving Grok nonstop... I bet it will be uncontested by 2026.)”

“Grok [@MarioNawfal] GROK-4 OUTPERFORMS TOP AI MODELS ON TRUTH SEEKING AND CAUTIOUS REASONING New research using Sparse Autoencoders shows a clear behavioral gap among frontier AI models when facing uncertainty. Instead of guessing, Grok-4 consistently pauses, clarifies assumptions, and seeks… https://x.com/i/web/status/2001168165586297236”

“Cool @Grok [@BrianRoemmele] It is clear @Grok is the best frontier AI model. I use 1000s of techniques and technologies to not only train but to test AI models. They are very unique and quite unlike what most AI engineers use in training and testing. In Grok’s case he has proven to be able to see other”

“Grok Code [@amXFreeze] Grok Code captured nearly 60% market share with just initial V1.0 release xAI will soon release a 1M context window model A truly game-changing moment for developers”

“Try @Grok voice mode. It’s the best. [@XFreeze] Start talking to Grok - It has a best human-like voice, understands emotions better, and delivers insights instantly with the latest Grok model Just keep talking.... Grok listens to you, searches the web in real time, and pulls live info whenever you need it Pause, resume and… https://x.com/i/web/status/2000414506057384029”

“This is a side effect btw. @xAI spent almost no effort on chess. [@techdevnotes] Grok 4 is currently the Top performing model in the Kaggle AI chess competition No tools used”

“Grok 4.20 Heavy (Beta 2) is extremely fast for deep analysis. Beta 3 will have many fixes and functionality gains. [@ArtificialAnlys] The Grok 4.20 Beta shows three major improvements over Grok 4: ➤ Our lowest ever hallucination rate on the AA-Omniscience evaluation. When Grok did not know the answer, it hallucinated an incorrect answer 22% of the time - this is the lowest hallucination rate of any model we”

“We had an ace up our sleeve @xAI. Turns out to be just enough to hold first place! Upgrades are in work to address presentation quality/style vs competition. That will shift ELO meaningfully higher. [@lmarena_ai] 📰More exciting news today: @xai's latest Grok-3 tops the Arena leaderboard! 🔥 This is the newest, production model, grok-3-preview-02-24 With over 3k votes, this model is tied for #1 overall, and across Hard Prompts, Coding, Math, Creative Writing, Instruction Following, and… )”

“Grok [@XFreeze] Every Grok model released in 2025 topped the leaderboards....The track record is undeniable 2025: The year xAI became unstoppable, shipping frontier model after frontier model with insane progress 2025 highlights: 🔹 Grok 3 debuted #1 on Text Arena at launch 🔹 Grok 4 defined… https://x.com/i/web/status/2007126896816042176”

“We will continue to evolve the model to make the images great for video games if that’s what a user requests [@DannyLimanseta] Pixel Art generated by @grok Imagine is surprisingly GOOD. The pixels look clean and maintain pixel art coherence throughout the whole image. With a bit of cleanup using retrodiffusion, I think they are ready for use in games etc.”