Large Language ModelText Generation

DeepSeek R1

DeepSeek's reasoning model trained with pure RL

Released 2025-01-20

Context Window

128K

API Access

Available

Benchmark Performance

reasoning

reasoning

Massive Multitask Language Understanding

90.8%

reasoning

Graduate-Level Google-Proof Q&A

71.5%

math

math

MATH

97.3%

coding

coding

HumanEval

85.0%

conversation

conversation

Chatbot Arena Elo

1358.0Elo