Openai charges $200/MO for reasoning tokens you never see.
Deepseek prints all 21.51 seconds of its thinking on screen for free, running on a board bitcoin miners threw away 00:32 the stats drop under the answer and he reads them off, "15.32 tokens per second, 392 tokens, 0.35
Top AI LLM models for every task
Ask Gemini 3.1 Pro for free right now 👉 → Writing & Research: GPT-5.4, Claude 4.6, Gemini 3.1 Pro, Perplexity → Social Content: Grok 4, GPT o3, DeepSeek → Academic / STEM: Claude Opus 4.6, MiniMax M2.7
Ollama v0.33 turns Claude Desktop into an interface that can run completely different models underneath it.
Local ones on your machine. Cloud ones on Ollama's infrastructure. Or back to Claude whenever you want. Setup on Mac: open Ollama, go to apps, find Claude, turn it on.
I use deepseek because it uses less water, it's Chinese
You have reached the end of the archive
All of deepseek