xv01

// > I have switched to Gemini 2.5 + openai + Qwen + deepseek

Anthropic don't have anything can balance their weakness (the nosense safety, and super expensive etc.)

gemini: super strong, fast and cheap openai: still have some magic nobody can beat qwen+ds: truly open Anthroic: ?

https://x.com/virushuo/status/1920119483529650238

======

https://www.ollama.com/assets/library/qwen3/37a622f9-71bb-49df-9fea-161b925e4853

qwen3:235b-a22b | qwen3:32b MoE Dense

— Aider Pass@2 || o1 2024-12-17 | DeepSeek-R1 | Gemini2.5-Pro | o3-mini Medium

https://www.ollama.com/assets/library/qwen3/7ac69e9c-c4d9-45c1-8bd6-a9787c8c454c

qwen3:30b-a3b | < qwq:32b > | qwen3:4b | gemma3:27b-it MoE Dense

— GPQA

** 4B parameter model ollama run qwen3:4b // 8b 14b

• 32B parameter model ollama run qwen3:32b https://www.ollama.com/library/qwen3:32b

** 30B mixture-of-experts model with 3B active parameters ollama run qwen3:30b-a3b

235B mixture-of-experts model with 22B active parameters ollama run qwen3:235b-a22b

cf. https://www.ollama.com/library/qwq

---

$ ollama list NAME ID SIZE MODIFIED qwen3:30b-a3b 2ee832bc15b5 18 GB 19 minutes ago qwen3:4b a383baf4993b 2.6 GB 39 minutes ago

======

• https://ollama.com/library/gemma3:27b

https://blog.google/technology/developers/gemma-3/

https://ai.google.dev/gemma/

======

• https://jina.ai/news/a-practical-guide-to-implementing-deepsearch-deepresearch/

======

https://api-docs.deepseek.com/

======

• ? https://ollama.com/library/openthinker:32b