Discussion

• @elonmusk Elon Musk on x

  Based on strong positive feedback from customers in our beta test program, @SpaceXAI will make Grok 4.5 available to the public tomorrow. It is an Opus-class model, but faster, more token-efficient and lower cost.

• @ericzakariasson Eric Zakariasson on x

  try grok 4.5 in cursor now! double usage for the first week included. its a good model [image]

• @zhmeishi Zhenmei Shi on x

  Have made Grok 4.5 + Grok Build as my agents for a while. They are excellent partners! Please try it and give us feedback 🚀

• @sawyermerritt Sawyer Merritt on x

  SpaceXAI has officially launched Grok 4.5 for the public. “Our first model trained specifically for coding and agents. It was trained with Cursor and offers frontier intelligence at leading speeds and cost efficiency.” [video]

• @forwarddeploy Umesh Khanna on x

  Been using grok 4.5 for everything from coding to drafting my emails/iMessages to even making a board game during our dinner parties 😂

• @jasonbud Jason Ginsberg on x

  please give Grok 4.5 a try. it's an incredibly useful daily driver.

• @mntruell Michael Truell on x

  Excited to release Grok 4.5 with @SpaceXAI. It's an Opus-class model that's fast and low cost. It's a significant step up over any model we've developed so far, including Composer 2.5, and has become the daily driver for many on our team. First of many releases. More soon.

• @robinebers @robinebers on x

  Cursor and Grok Build now have Grok 4.5 I've tested it since yesterday and it's a really solid model, big departure from Composer 2.5 weak on design, but *very* solid coder plus, faster + cheaper this is big! [image]

• @benjitaylor Benji Taylor on x

  Grok 4.5 is now available. Has been great to play with the model internally over the past few weeks. It's a real step change.

• @imsh4yy Shayan on x

  If the Grok 4.5 benchmarks hold, this is fantastic. Input: $2 / 1M Output: $6 / 1M [image]

• @cursor_ai @cursor_ai on x

  We've partnered with SpaceXAI to train Grok 4.5. It's our most powerful model yet and the first we've built for more than software engineering. [image]

• @spacexai @spacexai on x

  Grok 4.5 is efficient and fast. It uses half as many tokens per task, delivers higher tokens per second, and costs less than half as much as comparable models, at $2/M input and $6/M output tokens.

• @spacexai @spacexai on x

  Grok 4.5 is built for real-world engineering. It excels in large codebases and handles long-running tasks that span multiple repositories, hundreds of skills, and a variety of tools. [image]

• @kimmonismus @kimmonismus on x

  Seriously, what?! Did NOT expect grok 4.5 to perform THAT well. On par with GPT-5.5, cheaper than opus 4.8. Cursor deal paid off [image]

• @ai_for_success AshutoshShrivastava on x

  Grok 4.5 is rolling out. Holy smoke, SpaceXAI cooked well 🔥 Bring this to Cursor soon. I want to test it. [image]

• @mweinbach Max Weinbach on x

  Grok 4.5 is in Cursor already [image]

• @teksedge David Hendrickson on x

  Oh wow check out these specs for Grok 4.5. Pretty strong results! Maybe Elon was underplaying his new model. [image]

• @bridgemindai @bridgemindai on x

  New CursorBench results just dropped and Grok 4.5 is the story. #3 overall at 66.7%. Right behind Fable 5 Max at 70.5%. Now look at the cost column. Fable 5 Max: $17.32 per task Grok 4.5 High: $1.51 per task That is Fable level performance at roughly 1/10th the cost. And it [imag…

• @scaling01 @scaling01 on x

  Grok 4.5 Benchmarks [image]

• @daniellockyer Daniel Lockyer on x

  Grok 4.5 was just released in Cursor 👀 Price is $2/M input and $6/M output Fast mode is $4/M input and $18/M output [image]

• @fixlationai @fixlationai on x

  🚨 Grok 4.5 OUT NOW Grok 4.5 has been released, benchmarks show it performing similarly to Opus 4.8! First Grok model to use Cursor data 👀 [image]

• @synthwavedd Leo on x

  Grok 4.5 is out. First time I've been impressed by a Grok model! Congrats on the launch @SpaceXAI [image]

• @elonmusk Elon Musk on x

  Our internal assessment is that Grok 4.5 is roughly comparable to Opus 4.7, but much faster. The combination of capability, faster speed and lower cost is what makes it competitive. We are closing the loop on real-world usefulness, not benchmarks. Hardcore engineers at Tesla &

• @elonmusk Elon Musk on x

  We will continue to make refinements to the Grok Build harness and the 1.5T foundation model almost every day in response to user requests. The 2T model will finish training this month and be available to customers next month.

• @yunta_tsai Yun-Ta Tsai on x

  Has been my regular work horse for weeks. Impressive indeed.

• @arnabgho Arnab Ghosh on x

  7 years ago we built iSketchNFill (ICCV'19): interactive sketch→image GANs with Adobe Research. The code had quietly bit-rotted into oblivion. Grok 4.5 resurrected the entire thing. Faster than Opus, just as sharp. Old research code never has to die now.

• @quantumfon Jon Farland on x

  Never, ever give up. The customers I've worked with over the last few days have already seen significant improvements in a wide range of applications. This includes enterprise and consumer applications alike. I'm excited to see what you build!

• @yagiznizipli @yagiznizipli on x

  In my first try, Grok 4.5 found 10+ optimizations in oxlint, and almost doubled the performance of the linter.

• @gauravisnotme Gaurav on x

  One of the best use cases I have found this is generating CUDA kernels - either directly based on a problem statement or looking through Torch profiler traces and identifying opportunities for optimizations. The model is extremely good at going on in a loop where it collects a

• @synthwavedd Leo on x

  The new “Big 3” is officially Anthropic, OpenAI, and SpaceXAI (sorry but every time I type that new name out it looks silly lol) You had a good run DeepMind, but what a comeback for xAI. Who knows what they could do if they took back the compute they've given to Anthropic and

• @teortaxestex @teortaxestex on x

  Heroes literally saved Grok project

• @gavinsbaker Gavin Baker on x

  Grok 4.5. Pareto dominant for coding by the numbers. We will see on the all-important vibes. Instinct is the benchmarks are likely directionally accurate given the stated focus on real world utility. [image]

• @teortaxestex @teortaxestex on x

  Grok seems to get tired and start lazily cheating later into the session (247K/500K), or I just got unlucky in any case, the sheer speed and decisiveness of the model are amazing, this is what Gemini 3.5 Flash was supposed to be [image]

• @kimmonismus @kimmonismus on x

  Grok 4.5: $2 in / $6 out GPT 5.6: $5 in / $30 out Opus 4.8: $5 in / $25 out Grok 4.5 seriously is freaking good price-performance wise. Opus 4.8 level, Sonnet 5 pricing

• @teortaxestex @teortaxestex on x

  I believe that serving a model with higher margins off your compute as an AGI lab founder is cuckoldry. And frankly it's good enough that Elon could ask more, he's clearly redeeming the brand name here. xAI (sorry SpaceXAI) + Cursor = OP

• @billyuchenlin Bill Yuchen Lin on x

  It is a great model, sir! And it's getting better and better. I'm very proud to work with the amazing team on this model @SpaceXAI! Would love to hear what you will build with Grok! Critical feedback is always appreciated! 🚀

• @zephyr_z9 @zephyr_z9 on x

  Testing Grok 4.5 Won't be surprised if Elon cancels the Anthropic & Google agreements and gets back the compute xAI is bacc

• @theo @theo on x

  Just realized I was testing Grok 4.5 extensively yesterday lol Pretty damn good and REALLY well priced

• @wholemars @wholemars on x

  I asked Grok 4.5 and Claude Fable to create a simple todo app. Grok build finished in 104 seconds, Claude Fable took 557 seconds. Screenshots of the results below, GitHub repo in comments. [image]

• @zdhnarsil Dinghuai Zhang on x

  Excited to deliver Grok 4.5, with which we pushed RL scaling to the next level and achieve top-tier agentic coding performance. The past few months have been intense for me — driving fundamental algorithmic changes on both RL stability and capability, iterating days and nights

• @andrewcurran_ Andrew Curran on x

  Grok strong! Was Elon actually underselling for once? [image]