Qwen 3.8 - 27B MLX-Serve 4bit model, made the best version of the famous Pagoda test so far for me. Using @pidotdev as the coding agent, and MLX-Serve as the backend.
Introducing GLM-5.3: Built to Code. Ready for Cyber Defense.
- Top-tier coding and agentic capabilities, achieved through post-training on the 743B base model
- A major leap in cybersecurity, setting a new standard among open models
Tech Blog:
Introducing Gemini 3.7 Flash : )
- it is fast!
- 50% lower price than 3.6 flash (through end of year)
- strong intelligence increase in only ~3 weeks (thanks to some awesome algorithmic improvements)
- available in the API, AI Studio, Antigravity, and more!
Deepseek-V4-Pro-0813 doesn't seem as insanely impressive as I expected.
This seems to be a pattern with Chinese AI labs:
smaller models like Qwen 27b, Deepseek-V4-Flash, and GLM-5.2 perform ridiculously well for their size,
but maybe due to a lack of training compute, their larger models feel a bit unaligned.
As they secure more compute, they will definitely get better over time.