Releasing Your Own Jev
Post-train a 4B/8B/27B judge on your agent's traces. It beats Jev. 79.7% agreement with human labels vs Jev's 66.3% Beats DeepSeek-V4.1-Flash (763B) by 14 points 0.13s per step on a single GPU Completely open source: recipe, data, training, evals
Top-5 Best Value #LLM Models of the Day at UTC-16
| Model | #AAII | Price | | GLM 5.3 Flash | 42 | $0.07 | | GLM 5.3 | 45 | $0.33 | | DeepSeek V4.1 Flash | 40 | $0.10 | | Claude Sonnet 5.5 | 56 | $3.06 | | MiMo-V2.6-Pro | 46 | $0.32 |
This is f**king insane
Deepseek just open sourced its entire agent runtime 240k+ github stars in about 6 weeks. built on one idea: everything is a plugin. → model adapter → tool registry → session log → sandbox → interface → the agent loop itself all swappable. no
Finally, an openrouter for agent harnesses.
Devs just open-sourced a plug-and-play infrastructure layer that lets you run different agent harnesses through one interface. Including: → Codex → Claude Code → Hermes → DeepSeek Harness → System One, powered by JEV → 9+ more
You have reached the end of the archive
All of deepseek