Holy god, Im using Deepseek V4 Flash and the cache hits on opencode v2 are INSANE!!
Damn, and the hot reloads are sweet, it also has a great and I mean great API surface , it comes with a nice Opencode Skill, and it finally truly feels like it has much better understanding and
Every layer of the AI model stack shipped open in one week - from Qwen's 2.4 trillion parameter…
Every layer of the AI model stack shipped open in one week - from Qwen's 2.4 trillion parameter frontier to Liquid AI's vision model running on a phone in 3GB. #OpenWeights #DeepSeek Full edition and full episode:
The best AI LLM models for every task
Try Gemini 3.1 Pro for FREE now 👉 → Writing and research: GPT-5.4, Claude 4.6, Gemini 3.1 Pro, Perplexity → Content for social networks: Grok 4, GPT o3, DeepSeek → Academic/STEM field:
The setup friction is basically zero, which matters more than it sounds.
1,500 free uses out of the box, across GPT, Claude, Gemini, Kimi and Deepseek. To get something just type it in the prompt box. Try it out at
You have reached the end of the archive
All of deepseek