all the models — AI benchmark observatory
← Models

DeepSeek-V2.5

DeepSeek · open weight · deepseek-v2.5

DeepSeek-V2.5 is an upgraded version that combines DeepSeek-V2-Chat and DeepSeek-Coder-V2-Instruct, integrating general and coding abilities. It better aligns with human preferences and has been optimized in various aspects, including writing and instruction following.

GSM8kHumanEvalMMLUArena HardMATHAider

Benchmark scores

BenchmarkScore
GSM8k0.95
HumanEval0.89
MMLU0.80
Arena Hard0.76
MATH0.75
Aider0.72
SWE-Bench Verified0.17

Pricing

  • No provider pricing.

AA metrics

No Artificial Analysis link yet.

Arena Elo

  • No Arena snapshot linked.