DeepSeek V4 Pro is absurd.
It just tied the most powerful AI on Earth while running dramatically leaner. Terminal Bench score: → Claude Fable 5: 88.0 → DeepSeek V4 Pro: 87.9 A tenth of a point apart. At a fraction of the cost. Here's what's wild: → 1.6 trillion
Deepseek v4 flash 0731 running on 2 sparks with MiaAI_lab repo, idk last time i was this excited xd
We have released ``AI Ensemble'', a Windows application that compares answers from multiple AIs side by side, as OSS.
Bulk send one question to OpenAI / Claude / Gemini / Grok / DeepSeek / Kimi / Qwen / Mistral / Cohere. You can directly compare the differences between each AI without concluding into a single conclusion.
This paper will cut your token costs by 5X!
And end bloated reasoning chains forever! Right now, all top models (like Claude, DeepSeek R1, or OpenAI o1/o3) generate thousands of "thinking" words before giving you an answer. They literally think out loud: "Okay, where do I
You have reached the end of the archive
All of deepseek