Skip to content

Accept All すべて承認

Modal

25 stories

ModalToolsHitting a billion tokens per minute on one GPU by combining a query planner and an inference engine Maximizing perf on AI-SQL queries with the KV-optimal left-deep join

ModalToolsHow to serve trillions of tokens for trillion-parameter coding agents Learn how we optimized performance and efficiency serving the workload that is changing software engineering forever — and you can too.

ModalToolsProduct updates: Sandbox Sidecars, new models, a refreshed dashboard, and more Recent product updates from Modal and news from around the community.

ModalToolsModal is expanding in Europe with our new London office Modal is expanding, and hiring on all fronts across Europe.

ModalToolsHow Botika runs full-stack generative AI on Modal Botika is building the next generation of tooling for agentic e-commerce brands, backed by a fleet of custom models trained and served on Modal.

ModalToolsQwen3.8-2.4T-A95B now available on Modal Qwen3.8-2.4T-A95B by Alibaba, with a 1M token context window, is now available via Modal Auto Endpoints.

ModalToolsBringing serverless functions closer to the speed of wire Modal’s Function Call data path is now >50ms faster. Our new routing layer is geographically distributed, so you can further reduce your network overhead.

ModalToolsA note on the Hugging Face agent incident Hugging Face published a technical timeline of a recent agent intrusion. Modal's platform and isolation were not compromised in this incident.

ModalToolsKimi K3 by Moonshot now available on Modal Kimi K3, a 2.8 trillion parameter multimodal model by Moonshot, along with a custom-trained DFlash speculator, is now available on Modal.

ModalToolsDevin Outposts on Modal Devin, built by Cognition, is an AI software engineer: it plans, writes, tests, and ships code semi-autonomously. With Outposts, Devin can now run its work in Modal sandboxes.

ModalToolsScaling to 1 million concurrent sandboxes in seconds How (and why) we built a scheduling system that can scale to 1 million concurrent sandboxes (per workspace) in seconds.

ModalToolsInkling by Thinking Machines now available on Modal Inkling, a general-purpose multimodal model by Thinking Machines, along with a custom trained DFlash speculator, is now available on Modal.

ModalToolsModal is a computer What is Modal? A machine that runs programs of arithmetic & logic operations on information.

ModalToolsHow to price serverless GPUs To compare rates for serverless and reserved GPUs, look at your application's peak-to-average ratio.

ModalToolsMulti-token Residual Prediction One tiny module, two ways to win on Diffusion LMs

ModalToolsAnthropic integration with Modal brings scalable compute to Claude Science Announcing our integration with Claude Science, bringing Modal's elastic compute to researchers when they need it.

ModalToolsModal's serverless Servers A deep dive inside our new ultra-low-latency primitive.

ModalToolsAchieve state-of-the-art inference latencies with speculative decoding How Modal and Decagon worked together to cut inference latency - and you can too.

ModalToolsIntroducing Modal Auto Endpoints: Optimized inference you actually own LLM inference at SotA speeds and Modal quality, now available to everyone.

ModalToolsUnpacking sandbox startup latency: why started ≠ ready Building performant sandbox systems goes way beyond the initial container boot. Here, we unpack what that means, and discuss some tools to help you manage the entire lifecycle.

ModalToolsSpeculation Is All You Need Why we're all-in on speculative decoding.

ModalToolsProduct updates: VM Sandboxes, Lower latency routing, RBAC, and more Recent product updates and news from around the community.

ModalToolsMaking FlashAttention-4 faster for inference What part of "dtype = 'fp8', num_splits = 0, pack_gqa = True, q_stage = 1, page_size = 1" do you not understand?

ModalToolsReinforcement learning is an infrastructure problem What we've seen helping teams run Reinforcement Learning at scale on Modal. Plus an open-source library to skip the scaffolding.

On the calendar

予定 · Conferences and launch events, this week's stories from each corner of the industry, and what its press is covering — in English

On view

This week, from that corner the newest stories from the corner of the industry you picked

The wire their own posts, translated

How code moves in 2026 from the industry's own reports

Sources

Fetched every three hours, linked to the original

About

The name is the button.

An agent writes the code now, and a human thumb hovers over Accept all a hundred times a day. The honest question of this decade is what, exactly, you just accepted. So I read the diff.

Eight years as a backend engineer in Stockholm, back in Tokyo since 2023, and I still ship something small most nights, usually badly. My one rule: the only real review of a model is the thing you built with it. That is why there is no leaderboard on this site.

Around a hundred and forty sources come through — the labs' own posts, the papers, the tools, the runtimes and languages nobody is watching yet, the chip under the demo, and the newsletters people actually forward. What is moving gets counted so the week reads at a glance, and every few days I explain one mechanism in plain words.

A story here is a headline, a short excerpt, one image and the link to whoever wrote it. We never reproduce the post; the images belong to their authors. A post written in Japanese appears in Japanese, exactly as it was published, and the link goes to the original.

— Yuki Halvorsen, Tokyo