They never revealed the base model for Composer 1 as far as I can tell
@simonw
-
Vibe Coding: Useful Shorthand for Code Review and Quality
By
–
I use the term vibe coding in a daily basis and it's really useful – "here's a demo of that working, it's just vibe coded though" is a useful shorthand for "this needs a whole lot of attention and review before we ship it"
-
Agentic Engineering: Redefining Professional Software Development
By
–
I refuse to admit defeat! I've been pushing "agentic engineering" hard as a better term for what professional software engineers are doing with coding agents
-
Top researchers departure: quit or fired explanation
By
–
Did we ever get a conclusive answer as to if their top researchers quit or were fired?
-
4-bit Quantization Fixes Tool Calling Performance Issues
By
–
Dan found that the 2-bit quantization broke tool calling but upgrading to 4-bit (at 4.36 tokens/second) got that working
-
Comparing Tool Calling Performance Between Q4 and Q2 Models
By
–
Have you observed a meaningful difference between Q4 and Q2 either when it comes to tool calling? Would love to see how you measure that
-
OpenAI acquires Astral: uv, ruff, and Typer tools
By
–
Thoughts on OpenAI acquiring Astral and uv/ruff/ty
-
Quantization Impact on Model Quality and Expert Reduction
By
–
How confident are you with respect to the output quality given the 2-bit quantization and reducing experts from 10 to 4? Did you have a mechanism for measuring that?
-
Running 397B MoE Model on M3 Mac with Efficient Weight Streaming
By
–
Dan says he's got Qwen 3.5 397B-A17B – a 209GB on disk MoE model – running on an M3 Mac at ~5.7 tokens per second using only 5.5 GB of active memory (!) by quantizing and then streaming weights from SSD (at ~17GB/s), since MoE models only use a small subset of their weights for