The FREE AI MODELS board just added GPT-6, DeepSeek V4.1-Flash, GLM-5.3-Flash, Qwen3.8-Flash, MiniMax-M3, Hy3, and MiMo-V2.5
Cavoti “sextet”: Hy3, MiMo-V2.5, DeepSeek-V4-Flash-0731, Qwen3.8-Flash, GLM-5.3-Flash, MiniMax-M3 at $0 through Sep 25 UTC. OpenAI-compatible:
Seems the meteorological Department was serious!!
It’s raining here !
Deepseek v4.1 flash vs gemini 3.8 flash.
Same prompt, same job: 4 animated three.js scenes in one self-contained html - storm, galaxy, tornado, tesla coil. deepseek: 6 calls, 14m, $0.15 gemini: 12 calls, 32m, $1.25 gemini blew exposure on 4/5 first attempts. even had a whole
Made a tool with Deepseek 4.1 Flash.
This thing is a beast and is the best Flash model out there. Previous best one was GLM 5.3 Flash but this is a much more capable model in every single way; and through the API and the Deepseek harness I'm getting 270+ tok/s which is crazy
You have reached the end of the archive
All of deepseek