Daily AI Briefing
The editor's 6 picks from the 120 stories AIBriefs curated that day.
Get this briefing in your inbox every morning
Carbon ranks #1 on CursorBench 4.0, Perplexity's WANDR and Zapier's AutomationBench at more than 2X lower cost per task than the next best external model. Priced like Gemini 4 Argon at $2/M input and $10/M output tokens; a /swarm multi-agent preview ships in Google Antigravity.
Anthropic, PBC filed a Form S-1 for a Class A common stock offering and intends to list on the Nasdaq Global Select Market under the ticker "ANT"; share count and price range are not yet set. Latent Space reports 1,400% growth, 85% margin, $134B cash, and $417B in future AI spending disclosed in the filing.
The flagship of Alibaba's next-gen Qwen family accepts image and video inputs and processes up to a million tokens of context. Developers reach the preview via Alibaba's Token Plan subscription, with early access also in the Qwen app, Qoder, QwenWork and QwenStudio.
Agents powered by Gemini Carbon wrote over six million lines of Lean proofs to verify Signal's core encryption, a project done in four weeks. Human experts reviewed just 1% of the codebase; no new vulnerabilities were found.
MiniMax published the weights for H3-Image, which it says produces 4,096-by-4,096-pixel images in under 10 seconds and accepts up to 16 reference images per request. Weights are on Hugging Face; the model is also served via MiniMax's API, MiniMax Code, Hailuo AI, and Vercel AI Gateway.
Gemini Robotics 2.5 boosts success rates 30-100% and runs 20-30% faster than the prior version across 200+ tasks, per DeepMind. Apptronik's Apollo 2 humanoid hit a 96% success rate picking and packing car parts on Mercedes-Benz production lines.
Also that day
102 more stories that day — open the live feed.
Get tomorrow's AI brief in your inbox