Mistral released 2 new model upgrades with a vision encoder and 15% improvement on math and coding benchmarks:
– Magistral Small 1.2
– Magistral Medium 1.2 Rolling out to Le Chat as well
CODE
-
Mistral releases new model upgrades
By
–
-
OpenAI’s Self-Referential Code Generation and Dario’s Prediction
By
–
OpenAI:" I use codex to write 99% of my changes to codex. I have a goal of not typing a single line of code by hand next year :)" So Dario was right afterall?
-
Claude AI writes majority of Anthropic’s code, not layoffs
By
–
""70, 80, 90% of the code written in Anthropic is written by Claude. (…) "I said something like this 3 or 6 months ago, people think of it as falsified because they think of it as like we're going to fire 70 80 or 90% of the software engineers but what really happens is that… pic.twitter.com/fVL8aYvmPh
— Chubby♨️ (@kimmonismus) 18 septembre 2025""70, 80, 90% of the code written in Anthropic is written by Claude. (…) "I said something like this 3 or 6 months ago, people think of it as falsified because they think of it as like we're going to fire 70 80 or 90% of the software engineers but what really happens is that
-
Hugging Face Hub Quality of Life Improvement Announcement
By
–
New QoL improvement on the HF Hub! cc @simonw as you were interested in this IIRC
-
Supercharged Coding with GenAI: GitHub Copilot and ChatGPT Best Practices
By
–
Supercharged Coding with GenAI – From vibe coding to best practices using GitHub Copilot, ChatGPT, and OpenAI See this new book here: http://
amzn.to/3VBH2Su from @PacktDataML -
Grok Code Team Releases Beta with Daily Improvements
By
–
Great work by Grok Code team! This is just a beta release. It is improving almost every day.
-
Grok Code Gets Major Upgrades and 1M Context Window
By
–
Many major upgrades & fixes to Grok Code coming out next week, including 1M context window
-
PP-OCRv5: Compact Powerful Optical Character Recognition Model
By
–
Check out PP-OCRv5, our small yet powerful model specializing in OCR.
-
CodeLLM CLI Now Features Advanced Browser Use Capability
By
–
Browser Use Enabled On CodeLLM The CodeLLM CLI now has state of the art browser use capability! The coding agent will not only complete your task, it will also test it
-
Gemini DeepThink Wins Gold at ICPC2025 Programming Contest
By
–
Yet another important milestone for Gemini #DeepThink in achieving goldat #ICPC2025, the world's most prestigious college-level programming contest. Big congrats to everyone in this great achievement! Stay tuned for more to come from DeepThink 😉