I’m excited to share new research work from the Snowflake AI Research Team focused on advancing enterprise AI systems. Arctic-Text2SQL-R2 is a reasoning model designed for enterprise SQL generation. Trained on Snowflake-native data and optimized for real-world enterprise SQL workloads, the specialized model outperforms larger frontier models on difficult SQL benchmarks despite being 30–150x smaller than other high-performing models. To make specialized models like Arctic-Text2SQL-R2 practical at scale, the team also introduced ZoRRo (Zero Redundancy Rollouts), a set of optimizations that eliminate redundant computation in long-context RL workflows. ZoRRo accelerated RL training by up to 3.5x, reducing runtime from over five days to only 1.5 days. It also reduced memory consumption enough to support 3.2x longer context windows, enabling more efficient training on complex enterprise reasoning workloads. Together, this work demonstrates how the next wave of enterprise AI innovation will be driven by both stronger domain-specific models and more efficient training systems. Read more in the blog posts in the comments: Arctic-Text2SQL-R2: snowflake.com/blog/engineeri… ZoRRo: snowflake.com/blog/engineeri… @yao_zhewei @yuxionghe @samyamrb @jeffra45 @StasBekman

Conference didn't share the attendee list until I arrived. No problem. Pasted the CEO list into @SnowflakeDB Intelligence. 2 minutes later: customers, prospects, ARR — all mapped. This is what AI your own data looks like in practice. Not a demo. Not a deck. Just answers.

CoCo is quickly becoming a force multiplier for our customers and partners alike. They're not just experimenting. They’re transforming how work gets done. The speed, the productivity gains, the shift in how teams build—it’s real, and it’s happening now. Don't take my word for it: — Trent Foley of @letsevolv: Cortex Code has become the core infrastructure for how we scale and drive adoption. A single integration session executed 2,500 automated actions... that's 5-8x productivity. Weeks of manual development happening in hours. — Leading Automotive Dealership Group: We’ve been digging away with shovels for years, and now Snowflake just showed up with excavators. It’s easily the most practical and well-developed AI tool we’ve seen. — Vibhor Gupta of @Shelter_Ins: Cortex Code reduces friction in everyday data and AI development while maintaining the oversight we need in a regulated environment.

We’re seeing a clear shift in how teams build with AI, moving from isolated assistance to deeply integrated, agentic workflows. With @Snowflake's latest updates to Cortex Code, we’re making that shift tangible. ❄️Cortex Code is now generally available in Snowsight, with a persistent AI coding agent embedded directly in the data workflow. 💻Cortex Code CLI now supports Windows, expanding access for developers working across different environments. 🤖Agent Teams enable coordination of complex, multi-step tasks by running work in parallel. The result: faster iteration, tighter feedback loops, and the ability to take on significantly more ambitious data and AI workloads, without adding complexity. Read more in the blog post below: snowflake.com/en/blog/cortex…

The shift to agentic enterprises requires grounding in trusted data, strong governance, and seamless action. Project SnowWork brings this to life: autonomous agents for business users that respect controls, observe every step, and drive real results. Excited to see this evolve → snowflake.com/en/blog/agenti…

Thrilled to share Jacobi Forcing from Snowflake AI Research—transforming autoregressive LLMs into parallel decoders via progressive distillation on generated trajectories, unlocking up to 4x inference speedup with near-AR quality preserved. Trains models on Jacobi decoding trajectories with a progressive noise schedule, shifting AR models to efficient parallel decoders while retaining causal attention and KV-cache compatibility. Achieves 3.8× wall-clock speedup on coding/math benchmarks (e.g., HumanEval, GSM8K) with minimal performance loss. Introduces multi-block decoding and rejection recycling for 4.5× more tokens accepted per forward pass, outperforming diffusion LLMs by 7-53× in speed-quality tradeoff. No architectural changes or draft models needed—seamless integration with existing serving systems. Huge shoutout to the team: Lanxiang Hu, Siqi Kou, Yichao Fu, Tajana Rosing, Zhijie Deng, @samyamrb , @haozhangml , @yuxionghe Paper: arxiv.org/abs/2512.14681 Code: github.com/Snowflake-Labs/ja… #AI #LLMInference #MachineLearning

Today, I'm excited to launch my lifelong passion project, Grand Old Books!! 🚀 There are 1000s of beautiful novels of the past, not in English, locked up in old PDFs, with no physical copies left. We started with Indian texts and brought back 12 books in 6 languages with pictures and annotations. This is, and will always be, completely free. We can't let time wash away history. Please comment to let me know what book you'd like to see added.

Couldn't agree more— this is transforming data workflows in ways we only dreamed of. Game changer! ❄️

Snowflake is at the center of the enterprise AI revolution, and our Q3 results show the momentum. 📈 Product revenue up 29% YoY to $1.16B, with RPO at $7.88B (37% YoY). 💡 Snowflake Intelligence marks our fastest product adoption ever, helping @TSImagine_ , @Fanatics & @USABS over a thousand more customers harness agentic AI. 🤝 Expanding impact through partnerships with @AnthropicAI, @SAP, @awscloud, @Accenture, @Workday, @PalantirTech, @splunk & @UiPath. 🚀 370 Product launches YTD (35% YoY), a record 615 new customers, and 40K #SnowflakeWorldTour attendees (40% YoY). The best is yet to come. ❄️❄️❄️ investors.snowflake.com/news…

Alright, quite a few things wrt Snowflake AI Research at @NeurIPS in San Diego this week 1. [Expo Booth] Come and talk to us and get a Snowflake T-shirt and swag 2. [Meetup] Snowflake x FastVideo - fireside conversations, food, light drinks - Thursday, Dec 5 @ 5pm - RSVP’s going fast! luma.com/u2fznuuh 3. [Paper] SuffixDecoding: Extreme Speculative Decoding for Emerging AI Applications by Gabriele Oliaro - Friday, Dec. 6 @ 11am | Exhibit Hall C,D,E #816 - Learn more: snowflake.com/en/engineering… 4. [Workshop] Arctic Inference: Breaking the Speed Cost Tradeoff in LLM Serving by aurick.qiao@snowflake.com - Friday, Dec. 6 @ 6:25pm | Hard Rock Hotel - Register: sites.google.com/mila.quebec… 5. [Jobs] We are hiring: jobs.ashbyhq.com/snowflake/f… See you at the conference.

Have you ever wondered by how much is your MoE implementation slower than its dense equivalent - let's say Qwen3-Next-80B-A3B and we want to compare its performance to its 3B dense equivalent which doesn't exist. Well, just set `config.num_experts=0` and voila, you get the dense equivalent w/o coding anything. You just won't get the shared expert in Next, but it's 512 vs 1, so it's quite negligible. Just remember you'd have to adapt the number of tokens when comparing because compute per token will be different. Thanks to @samyamrb for this last insight since I originally completely missed it!

18 months ago, @karpathy set a challenge: "Can you take my 2h13m tokenizer video and translate [into] a book chapter". We've done it! It includes prose, code & key images. It's a great way to learn this key piece of how LLMs work. fast.ai/posts/2025-10-16-kar…

AI doesn’t steal jobs. It hands you the keys to all of them.

In era of pretraining, what mattered was internet text. You'd primarily want a large, diverse, high quality collection of internet documents to learn from. In era of supervised finetuning, it was conversations. Contract workers are hired to create answers for questions, a bit like what you'd see on Stack Overflow / Quora, or etc., but geared towards LLM use cases. Neither of the two above are going away (imo), but in this era of reinforcement learning, it is now environments. Unlike the above, they give the LLM an opportunity to actually interact - take actions, see outcomes, etc. This means you can hope to do a lot better than statistical expert imitation. And they can be used both for model training and evaluation. But just like before, the core problem now is needing a large, diverse, high quality set of environments, as exercises for the LLM to practice against. In some ways, I'm reminded of OpenAI's very first project (gym), which was exactly a framework hoping to build a large collection of environments in the same schema, but this was way before LLMs. So the environments were simple academic control tasks of the time, like cartpole, ATARI, etc. The @PrimeIntellect environments hub (and the `verifiers` repo on GitHub) builds the modernized version specifically targeting LLMs, and it's a great effort/idea. I pitched that someone build something like it earlier this year: x.com/karpathy/status/188467… Environments have the property that once the skeleton of the framework is in place, in principle the community / industry can parallelize across many different domains, which is exciting. Final thought - personally and long-term, I am bullish on environments and agentic interactions but I am bearish on reinforcement learning specifically. I think that reward functions are super sus, and I think humans don't use RL to learn (maybe they do for some motor tasks etc, but not intellectual problem solving tasks). Humans use different learning paradigms that are significantly more powerful and sample efficient and that haven't been properly invented and scaled yet, though early sketches and ideas exist (as just one example, the idea of "system prompt learning", moving the update to tokens/contexts not weights and optionally distilling to weights as a separate process a bit like sleep does).

We published new speculative decoding models for gpt-oss-20b and gpt-oss120b! They are based on the LSTMs and make gpt-oss generation 1.6-1.8x faster 🚀 The speculator models are open-sourced and ready-to-run in Arctic Inference 👇

Our team trained and released Arctic Speculator, which improves vllm generation speed by 1.6-1.8x for OpenAI’s recent gpt-oss models. Check it out here: snowflake.com/en/engineering…