AI Daily Brief — 09 March 2025
Sunday turned into Manus's first reality check. TechCrunch's hands-on review under the headline "probably isn't China's second DeepSeek moment" — testing four tasks that all failed…
Sunday turned into Manus's first reality check. TechCrunch's hands-on review under the headline "probably isn't China's second DeepSeek moment" — testing four tasks that all failed…
International Women's Day Saturday delivered the year's first major Western-built generalist agent. UK startup Convergence AI debuted Proxy 1.0 — claiming 88% on WebVoyager, ahead of…
Friday delivered Manus day-two — the viral hype peaked, then the technical reverse-engineering started. Victor Mustar called it "the most impressive AI tool I've ever tried."…
Thursday delivered the year's biggest Chinese agent debut, the most aggressive Western OCR drop, and the first credible RL-reasoning open-weights challenger to DeepSeek-R1. Manus AI launched.…
Wednesday delivered the year's most consequential research-recognition moment. ACM named Andrew G. Barto and Richard S. Sutton recipients of the 2024 A.M. Turing Award for developing…
QWEN CHAT Hugging Face ModelScope DEMO DISCORD Scaling Reinforcement Learning (RL) has the potential to enhance model performance beyond conventional pretraining and post-training methods. Recent studies…
Tuesday absorbed Monday's capital and infrastructure shock. Anthropic's $3.5B Series E at $61.5B post-money kept echoing through analyst coverage. CoreWeave's S-1 continued dominating AI-infra coverage. Vertex…
Monday delivered the year's biggest single-day capital and infrastructure cluster. Anthropic announced a $3.5B Series E at $61.5B post-money — Lightspeed-led. NVIDIA-backed CoreWeave filed S-1 ($1.9B…
Sunday was a genuine transition day. No dated frontier-lab launches. The week's narrative carried over from Claude 3.7 Sonnet (Feb 24) and GPT-4.5 (Feb 27) into…
Saturday closed DeepSeek's batched release week with a "One More Thing" surprise. The lab published a technical write-up of the V3/R1 inference system: 73.7k input /…
Friday closed both Open Source Week and February with a thoroughgoing infra finale. DeepSeek capped its five-day batched release with 3FS hitting 6.6 TiB/s on a…
Thursday delivered OpenAI's largest model ever and the most controversial pricing decision of the year. GPT-4.5 "Orion" research preview shipped at $75/$150 per 1M tokens —…
Wednesday delivered NVIDIA's biggest quarter ever and another Chinese open-source kernel drop. NVIDIA Q4 FY25: $39.3B revenue (+78% YoY), Blackwell $11B in first quarter — fastest…
Tuesday delivered a double Chinese open-source punch and Microsoft's biggest consumer-AI commoditization yet. DeepSeek opened Day 2 of Open Source Week with DeepEP. Alibaba simultaneously open-sourced…
What's Changed fix: do not use python_tag when encoding non-code_interpreter tool_calls by @ehhuang in #283 fix: tool_call was not encoded by @ehhuang in #284 Full Changelog:…
Monday delivered Anthropic's biggest model release since 3.5 Sonnet plus the year's most consequential open-source infrastructure drop. Claude 3.7 Sonnet shipped as the industry's first hybrid…
QWEN CHAT DISCORD This is a blog created by QwQ-Max-Preview. We hope you enjoy it! Introduction Okay, the user wants me to create a title and…
Sunday delivered xAI's voice-mode debut. Grok Voice early beta went live in the iOS app for X Premium+ and SuperGrok subscribers — Ara and Rex voices,…
Saturday delivered an xAI rebrand and a calm pre-Claude weekend. xAI updated the Grok brand mark to a black-hole singularity glyph paired with the new tagline…
Friday delivered OpenAI's biggest international agent rollout to date and DeepSeek's Open Source Week announcement. Operator expanded to UK, Australia, Brazil, Canada, India, Japan, Singapore, South…
Thursday delivered the year's most consequential humanoid-robotics release and xAI's free-tier opening. Figure AI introduced Helix — first VLA to output high-rate continuous control of the…
Wednesday delivered Microsoft's biggest Nature paper of the year and Mistral's strongest consumer milestone. Microsoft Research with Ninja Theory published Muse — the first World and…
Tuesday delivered a stealth-out from a former OpenAI CTO and the year's most provocative open-source post-train. Mira Murati unveiled Thinking Machines Lab. Perplexity shipped R1-1776 (DeepSeek-R1…
Monday delivered xAI's biggest swing of the year. Grok 3 + Grok 3 mini in livestream — 93.3% AIME 2025, 84.6% GPQA, 79.4% LiveCodeBench. "Chocolate" first…