
Ai2 and Hugging Face Expand Partnership, Unlocking 2 Petabytes for Open AI
Ai2 triples its Hugging Face storage to ~2 petabytes and drops rate limits, cementing its position as the Hub's most prolific open-science publisher
20 stories

Ai2 triples its Hugging Face storage to ~2 petabytes and drops rate limits, cementing its position as the Hub's most prolific open-science publisher

Alibaba's Wan3.0 enters public beta with 30-second native video generation, document-to-video inputs, and API pricing starting at $0.05/sec

SandAI released MAGI-2 Preview, a 114B open-source mixture-of-experts video model that activates only 6B parameters per token and generates 10-second 1080p clips with synchronized audio.

Warp's Agent CLI now lets you write plain-English routing rules that automatically send each prompt to the right model, shareable across your whole team

A community LoRA accelerates MiniMax-H3 joint video and stereo audio generation to as few as 4 sampling steps, roughly 5x faster than default.

Google's Creative Lab open-sources a fully offline, Raspberry Pi-powered voice translator built with Gemma 4, Moonshine ASR, and 3D-printed hardware

Browserbase open-sources a self-improving browser agent that writes its own playbooks, cutting per-run costs by 45% after just a few iterations

BabelDOC translates scientific PDFs into bilingual documents while preserving layout, formulas, and terminology — now backed by an ACL 2026 paper
If your IR path is paste-into-a-frontier-API, it will break when you need it most.

Prime Intellect's open-source Prime Agent hits 95.5% on ARC-AGI-3 by letting models rewrite their own scaffolding at runtime

Meta launches Muse Code, a terminal coding agent with persistent sub-agents and a co-trained model that can grind GPU kernels for 24 hours straight

Zed v1.14 ships OS-enforced sandboxing for its AI agent's terminal and fetch tools, blocking rogue writes and network calls by default

Goodfire's MAPS uses ESM-C 6B to explain 2.1 million protein variants, predicting not just pathogenicity but the specific biological mechanism disrupted

Hark's Handoff claims the #1 spot on Online-Mind2Web, beating GPT-5 and Claude Opus at a fraction of the token cost

Liquid AI and MacPaw are co-building a full on-device AI stack for Mac, with LFMs, local inference, and persistent memory — opening up to thousands of Setapp developers

Cursor launches a TypeScript SDK in public beta, letting teams embed its full coding agent runtime into CI/CD pipelines, internal tools, and customer-facing products.

Alibaba's Qwen team shipped a 27B open-weight multimodal model claiming Opus 4.6-class coding scores, runs on 17GB of VRAM under Apache 2.0.

Sarvam AI launches a formal partner network with AWS, HCLTech, IBM, HP, BCG, and others to scale India's sovereign AI stack into every enterprise

Vellum AI open-sources its personal assistant: 8 memory types, proactive reach-outs, and multi-channel presence across macOS, Telegram, and Slack

Databricks' Unity AI Gateway hits GA with hard spend caps, smart routing, and unified governance across agents, MCPs, and coding tools — over a quadrillion tokens already processed.