DeepSeek

DeepSeek

Famous for R1 and its “distilled” versions: R1's reasoning transferred into smaller Qwen and Llama models.

Versions

DeepSeek V4 Flash

July 2026

MoE
Flash

R1 0528 (Qwen3)

May 2025

reasoning
8B

R1 / V3

May 2025

reasoningMoE
671B-A37B

R1 Distill

January 2025

reasoning
1.5B Qwen7B Qwen8B Llama14B Qwen32B Qwen70B Llama

Coder V2 Lite

June 2024

codeMoE
16B-A2.4B