AI BriefingDaily AI briefing
Editor's picks come first, followed by stories repeatedly covered across days and sources.
Top stories
05- 01Anthropic: AI Models May Self-Replicate on IoT Devices by Late 2026
Anthropic's Frontier Red Team estimates that small AI models will reach the capability threshold for autonomous self-replication on commercial IoT devices by late 2026.
Anthropic Research
- 02Meta's Muse Wonder model now available on AI Gateway
The model features a 1M-token context window and supports text, image, and PDF inputs for coding and agent workflows.
Vercel Blog
- 03OpenBMB Releases MiniCPM-V 4.7 35B-A3B, First Open-Source MoE Vision Model with Smooth On-Device Inference
The new model achieves 13+ tokens/s decoding speed on iPhone 17 Pro Max using Lookahead Expert Scheduling.
Reddit
- 04Z.ai Launches GLM-Legal Beta: Matches Frontier Models on LegalBench and PLawBench
GLM-Legal achieves a 19.2% all-pass rate on Harvey LAB and parity with closed-source models on LegalBench and PLawBench while running on a leaner parameter footprint.
Hacker News
- 05xAI Launches Unified X Subscription Bundling Grok, Cursor, and X Premium
The new subscription tiers range from Free to $200 per month, offering shared AI usage pools across Grok and Cursor.
xAI
Developer tool releases
02Rising repositories and models
03- abenzerps/Qwen-Image-2.1-Uncensored-Quantized
Run Qwen-Image 2.1 Locally with 4GB VRAM
- microsoft/BitNet
Run 100B LLMs at Human Speed on a Single CPU
- baidu/Unlimited-OCR
OCR model that converts hundreds of PDF pages to text in a single pass