8x22 Instruct
β| 8x22 Instruct | https://models.mistralcdn.com/mixtral-8x22b-v0-3/mixtral-8x22B-Instruct-v0.3.tar | `471a02a6902706a2f1e44a693813855b` |β
What the smart people are recommending. 7867 books, tools, and products endorsed by the thinkers absorb.md tracks. Ranked by how many times each has been recommended across compiled podcasts, papers, posts, and tweets.
β| 8x22 Instruct | https://models.mistralcdn.com/mixtral-8x22b-v0-3/mixtral-8x22B-Instruct-v0.3.tar | `471a02a6902706a2f1e44a693813855b` |β
β| Pixtral 12B Base | mistralai/Pixtral-12B-Base-2409 | https://huggingface.co/mistralai/Pixtral-12B-Base-2409 |β
β| Codestral 22B | https://models.mistralcdn.com/codestral-22b-v0-1/codestral-22B-v0.1.tar | `1ea95d474a1d374b1d1b20a8e0159de3` |β
β| Mathstral 7B | https://models.mistralcdn.com/mathstral-7b-v0-1/mathstral-7B-v0.1.tar | `5f05443e94489c261462794b1016f10b` |β
β| Codestral-Mamba 7B | https://models.mistralcdn.com/codestral-mamba-7b-v0-1/codestral-mamba-7B-v0.1.tar | `d3993e4024d1395910c55db0d11db163` |β
β| Nemo Base | https://models.mistralcdn.com/mistral-nemo-2407/mistral-nemo-base-2407.tar | `c5d079ac4b55fc1ae35f51f0a3c0eb83` |β
β| Nemo Instruct | https://models.mistralcdn.com/mistral-nemo-2407/mistral-nemo-instruct-2407.tar | `296fbdf911cb88e6f0be74cd04827fe7` |β
β| Mistral Large 2 | https://models.mistralcdn.com/mistral-large-2407/mistral-large-instruct-2407.tar | `fc602155f9e39151fba81fcaab2fa7c4` |β
β| Pixtral Large Instruct | mistralai/Pixtral-Large-Instruct-2411 | https://huggingface.co/mistralai/Pixtral-Large-Instruct-2411 |β
β| Pixtral 12B | mistralai/Pixtral-12B-2409 | https://huggingface.co/mistralai/Pixtral-12B-2409 |β
βWe shipped loud, per-turn notifications in product for when requests are downgraded to the CLI v0.102.0. They are coming to other clients asap.β
β| Mistral Small 3.1 24B Instruct | mistralai/Mistral-Small-3.1-24B-Instruct-2503 | https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503 |β
βGPT-5.4 nano is is also available starting today in the API.β
β| Mistral Small 3.1 24B Base | mistralai/Mistral-Small-3.1-24B-Base-2503 | https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Base-2503β
βused try to make a new qmdresearcher directoryβ
βthis week on Upstream I wanted to share a conversation I had with burn Hobart author of The Tech and finance newsletter the diffβ
βqmd 2.0 is out. You can now use it in as a library with stable interface. Much easier to integrate. And all the other interfaces (cli, mcp, http) are now implemented in terms of that library, so that β¦β
βThanks to AI everybody can be a software engineer. In fact if you think you can't be you can go fire up Claude right now or any of your favorite chat bots and you can go start talking to it. You'd be β¦β
βI got a chance to play the game with Jeff, and it's incredibly beautiful (and fun). You can wishlist it on Steam now.β
βOne thing that's really changed over the course of 2025 was people started to use chat GBPT for much more personal very intimate applications.β
βamux ships as a [pi](https://github.com/mariozechner/pi) package with tools: - `amux_tail` β tail panel output with offset continuationβ
βWe show that quantum simulation of 2D FermiβHubbard model systems is a particularly well-suited application for the STAR architecture, requiring only hundreds of thousands of physical qubits and runtiβ¦β
βDetails: https://mistral.ai/news/mistral-ai-and-nvidia-partner-to-accelerate-open-frontier-modelsβ
βamux ships as a [pi](https://github.com/mariozechner/pi) package with tools: - `amux_shell` β run a command in a panel (wraps `run`)β
βAre you up for a challenge? https://openai.com/index/parameter-golf/β
βLearning algorithms can be quite complicated and today programming frameworks like TensorFlow, PyTorch, Cafe and many others can save you a lot of time.β
βWeβre building https://t.co/zwHPbMHK45 to make this universal interface as easy to deploy as possible.β
βJim has also the unique distinction of being OpenAI's very first intern back in 2016 where he co-developed World of Bits, and we're going to hear a little bit more about that later.β
βLast week I announced Context Hub (chub), an open CLI tool that gives coding agents up-to-date API documentation. ... Install chub as follows, and prompt your coding agent to use it: npm install -g @aβ¦β
βWe implement our approach using the MP-SPDZ framework.β
βFrom Lang Chain's early days as an open source framework to the broader evolution of Langraph, deep agents, Langmith, and agent builder, this episode is a deep dive into the frontier of the AI stack.β
β@Entropic_AI system looks good, but you need to figure out a path to run with 8gb M1,M2 sβ
βGet started here: https://www.perplexity.ai/enterprise/cometβ
βExperimenting with Starlette 1.0 with Claude skillsβ
βSubagents are now supported in Codex. They're very fun and make it possible to get large amounts of work done *quickly*:β
βyes, it is awesome! https://ai.google.dev/gemini-api/docs/maps-groundingβ
βSo, I built um I have a claw basically that takes care of my home and I call him Dobby the elf claw.β
βWe talked to Nick Tran, a nuclear engineer at TerraPower and writer of the excellent website whatisnuclear.comβ
βIf you're on Pro, it's particularly satisfying to ask Codex to spawn fast subagents using Spark!β
βyou've been on this site and I'm sure a lot of your your listeners have too, but this is the Google lab site. Yeah. As far as I know, it's the only Google site with a hot pink banner at the top, but iβ¦β
βBut maybe what I thought we could start with is Pomel. >> I don't know if you noticed this internet has been filled with videos about Pomel. It has. >> People have been loving it.β
βNoted. Okay. Two more for you. That's okay. Let's do a little Genie and then we can jump over to Gemini. So, um, Genie, I think, is one of those projects where you kind of have to see it to believe itβ¦β
βUm the new Gro 4.20 is it's really really good.β
βmore info on the benchmarks here: https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-flash-live/β
β@GoogleLabs: https://www.google.com/about/careers/applications/jobs/results?q=Google%20Labs&company=Googleβ
βNote: If you are on Chrome and have recently updated to version 142, you will need to disable the [`local-network-access` flag](https://developer.chrome.com/release-notes/142#local_network_access_restβ¦β
βThis example uses the [Apps SDK UI library](https://github.com/openai/apps-sdk-ui) for simple components such as images, buttons, and badges.β
βCheck out our blog post for details: https://mistral.ai/news/voxtral-ttsβ
βRead more: https://www.perplexity.ai/hub/blog/perplexity-apis-power-ai-browsing-on-samsung-devicesβ
βI mean, before Google, there was Alta Vista. There was CompuServe as online social kind of thing. Alta Vista was the search engine.β
βWeβve built an empirically validated, first-of-its-kind toolkit to measure AI manipulation in the real world β to better understand how it can occur and help protect people.β
βGpuer uses macOS system interfaces and command-line tools rather than private frameworks.β
βExperiment with Voxtral TTS directly in the Mistral Studio playground.β
βread the [docs](https://docs.mistral.ai) to learn how to use it.β
βThe SDK can be installed with either [npm](https://www.npmjs.com/), [pnpm](https://pnpm.io/), [bun](https://bun.sh/) or [yarn](https://classic.yarnpkg.com/en/) package managers.β
βGet your own Mistral API Key: <https://docs.mistral.ai/#api-access>β
βWe have dedicated SDKs for the following providers: - [GCP](https://github.com/mistralai/client-ts/tree/main/packages/mistralai-gcp)β
βSwiftUI menu bar app for monitoring macOS GPU and memory stats.β
βThis project is licensed under the [Apache License 2.0](LICENSE).β
βChatKit is a batteries-included framework for building high-quality, AI-powered chat experiences. Itβs designed for developers who want to add advanced conversational intelligence to their apps fastβwβ¦β
βSo you can go to tokenprobe.cchs.colia.edu. Is it still running? It's still running. It's still running and people come to it. Uh I use it in my classes uh to get students to do assignments.β
βYou know, this was before Product Hunt, before Techrunch, before I was the first one to link to Techrunch, the first one to link to Product Hunt, the first one to link to Tech Meme.β
βYou know, this was before Product Hunt, before Techrunch, before I was the first one to link to Techrunch, the first one to link to Product Hunt, the first one to link to Tech Meme.β
βYou know, this was before Product Hunt, before Techrunch, before I was the first one to link to Techrunch, the first one to link to Product Hunt, the first one to link to Tech Meme.β
βI mean, before Google, there was Alta Vista. There was CompuServe as online social kind of thing. Alta Vista was the search engine.β
βSpaces: A CLI Built for Humans and Agentsβ
βI also had a Microsoft Net Meeting site because they were com combined. They both work together.β
βpeople rarely need all the APIs supported in LiteLLM, might as well build a custom router with only what you need on the flyβ
βIn zero-shot custom voice tests, Voxtral TTS outperformed ElevenLabs v2.5 Flash - judged by native speakers for naturalness, accent accuracy, and similarity to the original voice.β
βLive per-process download/upload rates via `nettop`β
βConnection summaries come from `lsof -iTCP -n -P`β
βFor questions about the repo, I recommend either using [DeepWiki](https://deepwiki.com/karpathy/nanochat) from Devin/Cognition to ask questions about the repo, or use the [Discussions tab](https://gitβ¦β
βFor questions about the repo, I recommend either using [DeepWiki](https://deepwiki.com/karpathy/nanochat) from Devin/Cognition to ask questions about the repo, or use the [Discussions tab](https://gitβ¦β
βSo we decided to build Stripe Projects to help agents instantly provision services from the CLI.β
βApps SDK UI is a lightweight, accessible design system for building high-quality ChatGPT apps β
βApps SDK UI requires **React 18 or 19** and **Tailwind 4**. - React: https://react.dev/learn/installation - Tailwind 4: https://tailwindcss.com/docs/installationβ
βTry Extract v2 today ββ
βText to Video https://replicate.com/wan-video/wan-2.7-t2vβ
βπ Nova-3 + Nova-3 Multilingual: production transcription for noisy, real-world, and multilingual audioβ
βπ Nova-3 + Nova-3 Multilingual: production transcription for noisy, real-world, and multilingual audioβ
βπ Aura-2: enterprise TTS with sub-200ms TTFB for clear, structured speechβ
βImage to Video https://replicate.com/wan-video/wan-2.7-i2vβ
βan OWL 2 ontology that formalises the semantics of AnIML and aligns it with the Allotrope Data Format to support future cross-system and cross-lab interoperability.β
βNew in NotebookLM: The Science Of Ben Franklin, a first-of-its-kind featured notebookβ
βSo, our our Google AI Studio X account is where we're posting a lot of the new features that we're putting out there. Anything guides go out there as well. Um so, that all goes there.β
βIf you have NPM package axios in your dependencies you need to make sure it's pinned to a known safe version, sounds like there's another supply chain attack in playβ
βyou basically have Opus 46, you have uh codeex 53 and essentially you get this shift where I think the the the tools and the foundational models were pretty good at writing code especially for new venβ¦β
βAnd Aura is like a hack, right? It's basically a hack to get beautiful design for people who are not designers. So, you know you're going to get you're about to get way more than 500 users because, yoβ¦β
βTry to use the the the prom builder. It doesn't matter if it's from Visero or not. You can even use the prom v uh the promp builder from my site and just learn about all of this and then go to your faβ¦β
β[**Download ac-tracer.zip**](https://github.com/tobi/ac-tracer/releases/latest/download/ac-tracer.zip) and drag & drop it onto Content Manager.β
βplace it inside a ChatGPT agent offeringβ
βTry to find the best designs out there either from Instagram from dribble, from Twitter and you know follow these people, collect these resources, collect your favorite design. Let me try to refresh tβ¦β
βMr. Chatterbox is a (weak) Victorian-era ethically trained model you can run on your own computerβ
βLast week, the White House proposed a national legislative framework for AI. A key component is a federal preemption framework to prevent a patchwork of state regulations that hamper AI development. Iβ¦β
βSo, Moneybot, which we'd like to think of as like a CFO in your pocket, but it's essentially like a proactive um a proactive uh chatbot that can take actions on your behalf within Cash App. That's buiβ¦β
βAlso, uh, check out askjoo.ai. You, uh, ask questions and an AI version of me will answer.β
βI know that the Sierra Club has been part of a group that put together something called the big tech unchecked toolkit. Um can you talk a little bit about what that is and and why it's a toolkit? Who β¦β
βManager bot, which is roughly a similar thing on the Square side, that's built on top of Goose.β
βInstall this plugin in the same environment as LLM. ```bash llm install llm-mrchatterboxβ
βmove even more resources to the OpenAI agent platformβ