Unsloth AI @unsloth.ai 路 Apr 24

DeepSeek releases DeepSeek-V4. 馃悑 - DeepSeek-V4-Pro: 1.6T params - DeepSeek-V4-Flash: 284B params DeepSeek-V4-Pro rivals Claude-Opus-4.6-Max, GPT-5.4-xHigh and Gemini-3.1-Pro-High. They support 1M context length, thinking and set new records for Codeforces.

43 likes 3 replies

?

Replies

Hakan Damar 路 Apr 26

The DeepSeek V4 series is a waste of time. They probably released it too early. The Qwen3.6-35B or 27B models look much better than the DeepSeek V4 Pro at this stage.

Bret 路 Apr 26

Really wanting to compare V4 Flash to Minimax 2.7 in a 2bit quant as that's as big as I could currently dream of running locally

Money Is Fake 路 Apr 24

Yeah, but can it play Doom?