// COMPARISON

DeepSeek-V2.5

A head-to-head benchmark comparison of DeepSeek-V2.5 across 7 evaluations.

The models

Benchmark comparison

Benchmark DeepSeek-V2.5
code
HumanEval 89%
LiveCodeBench 41.8%
SWE-bench Verified 16.8%
coding
Aider Benchmark 72.2
general
AlignBench v1.1 8.04
AlpacaEval 2 50.5%
ArenaHard 76.2%

Best result per row highlighted in cyan. Each benchmark links to its definition and sources; each model links to its full scorecard.

Frequently asked questions

Which of these models is best for coding? ▶

DeepSeek-V2.5 has the strongest result on HumanEval among the models compared here. See the code-category rows in the table for the full picture.

How many benchmarks are compared? ▶

This comparison covers 7 benchmarks on which at least one of the selected models has a published score.

Where do the benchmark scores come from? ▶

Scores are aggregated from official model cards, technical reports and standard public evaluations, and link back to each benchmark's source. They are updated as new results are published.

Explore further

→ DeepSeek-V2.5 → AI benchmarks leaderboard → Build a different comparison
Build a different comparison

Models (0 selected)

Benchmarks (auto)

Select at least two models.