Bryce Hunt

@AmcaInc only 1 person could have done this...@ryanli28

First day at our new HQ ✅

@cognition Game changer. Fast is ridiculously fast

We’re releasing SWE-1.6, our best model in both intelligence & model UX. SWE-1.6 matches our Preview model on SWE-Bench Pro while dramatically improving on various behavioral axes. It’s available today in Windsurf in two modes: free tier (200 tok/s) and fast tier (950 tok/s).

Thanks for having us! was a blast

@sharedalbums @AlexanderMclane @zodichalat @Tiffanypianist @philhchen @JoanneZPeng @cocohearts @CarlSchoeller @silasalberti 🫵

FM18: San Francisco @AlexanderMclane @zodichalat @Tiffanypianist @philhchen @JoanneZPeng @bryceehunt @cocohearts @CarlSchoeller @silasalberti

@sharedalbums @bernhardsson @breeves08 @malharhar @ypatil125 @dalvabaird Best one yet

FM17: San Francisco @bernhardsson @breeves08 @malharhar @ypatil125 @bryceehunt @dalvabaird

@cognition Self testing feels like magic 🚀🚀

Introducing Devin 2.2 – the autonomous agent that can test with computer use, self-verify, and auto-fix its work. Try it for free! We’ve also overhauled Devin from the ground up: - 3x faster startup - fully redesigned interface - computer use + virtual desktop ...and hundreds more UX and functionality improvements.

On DeepWiki and increasing malleability of software. This starts as partially a post on appreciation to DeepWiki, which I routinely find very useful and I think more people would find useful to know about. I went through a few iterations of use: Their first feature was that it auto-builds wiki pages for github repos (e.g. nanochat here) with quick Q&A: deepwiki.com/karpathy/nanoc… Just swap "github" to "deepwiki" in the URL for any repo and you can instantly Q&A against it. For example, yesterday I was curious about "how does torchao implement fp8 training?". I find that in *many* cases, library docs can be spotty and outdated and bad, but directly asking questions to the code via DeepWiki works very well. The code is the source of truth and LLMs are increasingly able to understand it. But then I realized that in many cases it's even a lot more powerful not being the direct (human) consumer of this information/functionality, but giving your agent access to DeepWiki via MCP. So e.g. yesterday I faced some annoyances with using torchao library for fp8 training and I had the suspicion that the whole thing really shouldn't be that complicated (wait shouldn't this be a Function like Linear except with a few extra casts and 3 calls to torch._scaled_mm?) so I tried: "Use DeepWiki MCP and Github CLI to look at how torchao implements fp8 training. Is it possible to 'rip out' the functionality? Implement nanochat/fp8.py that has identical API but is fully self-contained" Claude went off for 5 minutes and came back with 150 lines of clean code that worked out of the box, with tests proving equivalent results, which allowed me to delete torchao as repo dependency, and for some reason I still don't fully understand (I think it has to do with internals of torch compile) - this simple version runs 3% faster. The agent also found a lot of tiny implementation details that actually do matter, that I may have naively missed otherwise and that would have been very hard for maintainers to keep docs about. Tricks around numerics, dtypes, autocast, meta device, torch compile interactions so I learned a lot from the process too. So this is now the default fp8 training implementation for nanochat github.com/karpathy/nanoc… Anyway TLDR I find this combo of DeepWiki MCP + GitHub CLI is quite powerful to "rip out" any specific functionality from any github repo and target it for the very specific use case that you have in mind, and it actually kind of works now in some cases. Maybe you don't download, configure and take dependency on a giant monolithic library, maybe you point your agent at it and rip out the exact part you need. Maybe this informs how we write software more generally to actively encourage this workflow - e.g. building more "bacterial code", code that is less tangled, more self-contained, more dependency-free, more stateless, much easier to rip out from the repo (x.com/karpathy/statu…) There's obvious downsides and risks to this, but it is fundamentally a new option that was not possible or economical before (it would have cost too much time) but now with agents, it is. Software might become a lot more fluid and malleable. "Libraries are over, LLMs are the new compiler" :). And does your project really need its 100MB of dependencies?

Introducing Arena Mode in Windsurf: One prompt. Two models. Your vote. Benchmarks don't reflect real-world coding quality. The best model for you depends on your codebase and stack. So we made real-world coding the benchmark. Free for the next week. May the best model win.

Meet Devin Review: a reimagined interface for understanding complex PRs. Code review tools today don’t actually make it easier to read code. Devin Review builds your comprehension and helps you stop slop. Try without an account: devinreview.com More below 👇

@michejafang great read

In 2025 | visited 12 factories in Shenzhen, hosted dinners for White House Al advisors, and met with dozens of robotics + hardware founders for Lux. 2026 will be turning point. This letter covers what surprised and intrigued me about this messy love triangle of SF, China, and Washington.

@sharedalbums hell yea

14 dinners got me everything I ever wanted in this world. This is a love letter to everyone who came to a Family Meal in the last 12 months. You took a chance on someone who wasn’t the traditional fit for this community, walked into dinners blind (a hidden guest list and no guarantee of what you’d get out of it) and gave me the one thing in this world you can’t buy back: your time. Growing up I was drawn to lot of startup movies/shows (The Social Network, Silicon Valley) and all I ever wanted as someone 10,000 miles away from where all of that was happening, was to be in it. In the restaurant industry, family meal is a moment when bartenders, servers, cooks, dishwashers eat together and get to know each other outside of service. Too often as founders, researchers, investors, engineers, you’re expected to always have the right answers, be “impressive” and in a constant state of being “on.” In a world where everything is so transactional, Family Meal was created to take you off your proverbial line and give everyone a break from the performance. For 14 dinners I’ve kept Family Meals sponsor-free to keep the intention behind it all pure. I didn’t want homework from anyone to deliver insights, founders, or potential recruits on a platter. In our world, money is a commodity and protecting what makes these dinners feel genuine is a price I’d happily pay. I think it has served us well. And through this experiment I’ve found some of the most non-transactional, inspiring, and quietly brilliant people that put me in awe everyday. 14 dinners got me everything I ever wanted in this world, and it has been a great pleasure being your host. Season 2 starts in 2026. your host, alicia

@mollycantillon @sharedalbums @breeves08 @f_x_wang @jacobtpl @rachelsupark @igarciacamargo @sarahzengy @philip_bogdanov @annarmonaco @qamcintyre @katieruthmishra @cory are we not??

@sharedalbums @breeves08 @f_x_wang @jacobtpl @rachelsupark @igarciacamargo @sarahzengy @philip_bogdanov @annarmonaco @qamcintyre @katieruthmishra @cory @bryceehunt we could be related @bryceehunt

FM14: San Francisco (Season Finale) @breeves08 @f_x_wang @jacobtpl @rachelsupark @igarciacamargo @sarahzengy @philip_bogdanov @annarmonaco @qamcintyre @katieruthmishra @cory @mollycantillon @bryceehunt

@velvetatom 🚀

Spent three years looking for a team in Biodefense to invest in. But never found one. So we built it ourselves. Valthos builds next-generation biodefense. As AI and biotech rapidly advance, we're approaching near-universal access to tools with the potential to cure humanity or make biological weapons. We have a short period of time to ensure that our capabilities to protect advance faster than the threats we face. At Valthos, we apply frontier methods to identify threats and compress the time from a biological sequence to validated medical countermeasures. Excited to share more with you all soon. Grateful to start Valthos with $30M in backing from @OpenAI @Lux_Capital @foundersfund, and many others incl. @Definition_Cap @topology_vc We are actively hiring - if you know engineers interested in solving this mission, reach out contact@valthos.com.

@andrei_serban @ThriveCapital @scale_AI @tryramp @webflow Congrats on launch!

All companies run on IT. But almost all IT teams are underwater. The ones that aren't run on Console. Console raised a $23M Series A led by DST Global Partners and @ThriveCapital to help leading companies like @scale_ai, @tryramp, and @webflow automate 50%+ of their tickets.

We’ve raised over $400M at a $10.2B post-money valuation to advance the frontier of AI coding agents. The round was led by Founders Fund with other existing investors including Lux, 8VC, Neo, Elad Gil, Definition Capital, and Swish VC all doubling down. We’re also joined by new investors including Bain Capital Ventures and D1 Capital. Two of our early investors, Christian Lawless of Conversion Capital and Emily Cohen of Neo, have even joined our team full-time.

@windsurf 👋🌊

Wave 12 is here, and it’s a big one! 📚 DeepWiki-powered docs for every symbol in your codebase 🔍 Vibe and Replace 🐛 100+ bugs squashed 🎨 Brand new UI … and more! Everything that’s new 🧵

@cognition @OpenAI 🚀

We’ve been working closely with the @OpenAI team to integrate GPT-5 into Devin. Starting today, you can select a preview version of Devin that uses GPT-5 as part of our agent orchestration. GPT-5 eval results 👇