jev isn't production ready. today @modaicdev is launching Mo, the first multimodal decision model that learns from its mistakes. most production tasks involve subjective decision-making, where the last mile of calibration and alignment has always fallen on the developer. Mo closes that gap on its own. it refines its instructions and weights from the signals already in your workflow to learn what a good decision means for you. use Mo for the decisions where alpha matters the most like LLM judging, lead scoring and qualification, content moderation, resume screening, and more. if you're already using another decision model like jev, plug it into Modaic and it becomes self-improving too :) 🐙 modaic.dev docs.modaic.dev

Just hit one year on the best infra team and they got Starter Bakery pastries to celebrate. Time flies when you’re having fun!!

@fal's H3 Max generates video faster than you can watch it. With director mode and voice prompting, you can direct a scene as it plays - moving the camera and guiding the action just by talking to the model. The leap isn’t just speed. It’s creative control. That’s what takes AI from impressive demos to a serious technology for Hollywood and professional filmmakers. Inspiring convo with @gorkem and @isidentical on what possibilities are unfolding in generative media.

Arrow 2 beats many big models on design and vector graphics. Super impressive

We’re partnering with @inferact (the team behind @vllm_project) to make TPUs a first-class citizen for open-source AI inference. Together, we’re delivering: - Upstreamed open-source TPU kernels - Native PyTorch integration via TorchTPU + so much more ↓

Vals AI co-founder and CEO Rayan Krishnan with a16z's Ben Horowitz and Jennifer Li on grading AI, what it costs, and who gets to make the rules: Every big industry eventually grows an independent testing layer. AI has credit ratings to learn from and Enron to avoid. Model capability today is still mostly self-reported. As public benchmarks saturate and models get better at optimizing for the tests themselves, Rayan makes the case for independent, continuously evolving evaluations. The harder problem is geopolitical. Reagan's "trust but verify" worked during the Cold War because you could fly over and count the missiles. No simple equivalent for AI models exists. In this conversation with Erik Torenberg, they get into how you measure a model's ability to improve itself, why every good benchmark eventually has to be retired, and what happens when token spend begins to rival employee salaries. 00:00 Intro 02:20 Llama 4 on public vs private benchmarks 05:24 Nobody agreed how to test humans either 06:55 What movie ratings teach us about AI 08:55 The Enron problem in benchmarking 11:36 Why a good benchmark has to be retired 13:22 Evals that run for weeks, not seconds 16:20 Where the real workday starts at 4pm 18:08 A firm really is just its evals 20:35 Why Sonnet can cost more than Opus 22:42 One engineer, 6 billion tokens in a day 25:05 Who should set the rules for models 28:55 Public sector enforces, private verifies 33:32 Why sovereign AI is inefficient and happening anyway 35:00 The AI version of trust-but-verify 37:15 Where cyber evals have to go next YouTube: piped.video/watch?v=WO9c9qxD… @RayanKrishnan @ValsAI @bhorowitz @JenniferHli @eriktorenberg

A very interesting breakdown that points to the future of data agents! Model capability improvements will continue to close gaps, but human guided semantic context for agents is still crucial for real world data messiness.

OpenAI's Mark Sellke and Mehtaab Sawhney with a16z's Lisha Li, on the state of AI and mathematics: Before OpenAI released GPT‑6 Astra last week, the model was already doing original mathematics. Recorded before the launch, this conversation tells the story of how it got there. It began with GPT‑5. Mehtaab Sawhney pasted in an Erdős problem still listed as open. Five minutes later, the model surfaced a paper that had solved it. The exercise eventually uncovered published solutions to 10 more problems thought to be open. Then Astra went further. Told to "go have fun" with a high-dimensional sphere-packing problem, it improved a bound that had stood since the 1970s. Mehtaab had spent six months on the same problem in graduate school and made "absolutely zero progress." Another Astra result established that non-sofic groups exist with a roughly 15-page proof. A related human breakthrough took 250 pages and machinery from quantum complexity theory. OpenAI’s Mark Sellke and Mehtaab Sawhney join a16z’s Lisha Li on why wrong ideas pollute a human’s context window, why polished papers hide how mathematics is actually made, why a breakthrough can stop one prompt early, and what math rewards once proving stops being the bottleneck. 00:00 Intro 02:44 Cracking an Erdős problem in 5 minutes 06:20 Why a human quits and a model doesn't 08:50 Wrong ideas pollute your context window 11:45 Why math papers are bad training data 16:20 Nobody knows how to stack spheres in high dimensions 18:28 The orange-stacking proof 21:04 The 1970s Russian paper nobody could beat 24:14 The function that won a Fields Medal 27:09 How Astra beat the sphere-stacking record 29:20 Why error correction is sphere packing in disguise 35:15 The breakthrough Astra almost didn't bother with 39:22 Solving harder problems means it has better taste 41:00 One model for taste, one for the grind 44:40 Astra found an infinite group no finite one can imitate 52:00 250 pages of quantum complexity, or 15 of group theory 56:18 Only humans write 200-page proofs 1:00:38 What changes when proving stops being the bottleneck 1:02:44 The problems AI may never solve YouTube: piped.video/watch?v=1JvyLGd2… @mehtaab_sawhney @MarkSellke @OpenAI @lishali88

With the pace of shipping in consumer AI and personal assistants, it's easy to forget how far we've come in terms of capability and autonomy. When chatGPT first came out it was an incredibly mind-opening moment for the general public - the biggest public exposure to LLM capability at the time. It was impressive in its conversationalness and writing, but also confident in its hallucinations and its knowledge stopped in 2021. Asking something as simple as the score to a recent game quickly outlined the gaps. Early chatGPT was magical, but the trust was not there for many, if not all tasks. Fast forward a few years and AI teammates/assistants like Grok @bot, Instinct, and others are redefining how we interact with AI systems and how autonomous they can be. Iterating on the momentum of OpenClaw and abstracting out and simplifying a lot of the core workflows, AI can do more, effectively. The biggest thing is a huge step function increase in trust. We now trust these systems to code, write, book, message, and act on behalf of us. If this is how far we've come in a few years, the next decade will be truly incredible.

Inference is one of the most interesting systems problems at scale. Near unlimited demand, wildly different workloads. Constantly shifting targets for models to optimize, and trade offs between cost, latency, throughput.

We're thrilled to lead a $300M investment in Gimlet Labs. AI inference is one of the fastest-growing markets in the history of capitalism, and we are running out of nearly every physical input required to serve it. Inference demand compounds at software speed; power plants, data centers, and semiconductor fabs do not. We believe @gimletlabs has built the solution: the first multi-silicon inference cloud, designed to produce more intelligence from every watt. This system is already delivering up to 10x gains in throughput and interactivity on frontier models within the same power envelope. In a market starved for compute, efficiency is net-new capacity and latency is product differentiation. Wherever the existing stack ends, the Gimlet team starts. Welcome, @zainasgar, Michelle Nguyen, @oazizi, @nserrino, James Bartlett, and the Gimlet Labs team! By @RaghuRaghuram, @sarahdingwang, @shangdaxu, @steph_zhang

Super insightful convo with @lishali88 and Daniel Litt!

Our fifth growth fund is now $8.5b. The founder is the asset in late stage, and we back the best in the world. At the same time, the best founders face global complexity earlier in their lives than ever before. So we are building out an enhanced Growth Platform for these exceptional founders. Biggest product cycles of our lives. Never been higher stakes, and never been more excited. @a16z

Machine Age!

Super excited for our new Machine Age Fund, which will invest in all of the computer infrastructure on which AI runs - including chips, memory, networking, and storage. And also the full systems AI runs on - from data centers to robots. This is a once in a generation opportunity to invest in improving the bottlenecks around hardware!

Distributed AI infrastructure works because of the network. GPU clusters rely on multiple network fabrics, each adding complexity as infrastructure scales, tenants change, and new capacity comes online. Traditional network automation wasn’t built for this dynamic environment or the hard tenant isolation it requires. That’s where NAAM – Network Automation, Abstraction, and Multi-Tenancy – comes in. As @a16z’s @martin_casado puts it: “The future of AI is gated on our ability to scale up the network. It is the harder problem.” In this video, Martin and a16z partners @appenz and @RaghuRaghuram join Netris CEO @alex_saroyan and co-founders Tigran Martirosyan and Arsen Arakelyan to discuss what’s changing in AI networking and why a new approach is needed. Netris is already putting NAAM to work across 35+ live AI deployments. Watch the video to learn why a16z believes the network is becoming one of AI infrastructure’s most critical challenges.

Honored to cohost another super fun event with the one and only Cursor & SpaceX team!! Sign up below 🙌🙌

Said it before and saying it again. It always takes a village to make a company, and to make a legend. Very proud of what we’ve built here - no one tries to be the only hero on a company or a deal, everyone works together to help our companies succeed. Go Team Infra!!! See how our team mom @KatieBaynes prepared us for a family photo.

The team in question :)