The fifth large-scale model test on WorkBuddy has one Kimi K3 left on WorkBuddy
Current conclusion: GLM-5.3 > GLM-5.3 Flash > Hy4 preview ≈ Kimi K3 > Hy3 > DeepSeek V4 Flash > DeepSeek V4 Pro prompt: Generate an svg animation: a pelican rides a bicycle, show it to me in H5
If you use Codex in your work, it opens up a new, very important option for you.
The Responses API now supports the DeepSeek model series, which means developers can run DeepSeek directly inside Codex using the API Key and Base URL.
TP=3 just isn’t too worth for GB10’s imo.
Granted I haven’t made recipes, but gains are small. There are better mixes of models for 1x spark and 2x sparks and a big jump in performance to 4x. Recommendations: Single = Qwen3.6 35b-a3b Dual = Deepseek V4-FV Quad = Deepseek V4-FV
Give Deepseek Harness the speech plugin it deserves.
Dsh plugin --profile web add allmodels/dsh-speech
You have reached the end of the archive
All of deepseek