// COMPARISON

DeepSeek-V3.1-Terminus

A head-to-head benchmark comparison of DeepSeek-V3.1-Terminus across 31 evaluations.

The models

Benchmark comparison

Benchmark DeepSeek-V3.1-Terminus
code
Codeforces Elo 2,073
FullStackBench 55.48%
LiveCodeBench 48.02%
LiveCodeBench v6 75.33%
MBPP 90.69%
coding
Aider Benchmark 92.86
MultiPL-E 77.68
general
ArenaHard v2 63.24
C-Eval 91.76%
IFEval 89.09%
MMLU-Pro 85%
MMLU-Redux 92.37%
MultiChallenge 45.79%
knowledge
AGIEval 89.83
CMMLU 89.2
TriviaQA 82.77
math
AIME 2024 71.67%
AIME 2025 89.06%
CNMO 2024 85.42%
HMMT February 2025 86.1%
Omni-MATH 81.93
UGMathBench 77.19
reasoning
ARC-AGI-1 40.62
BBEH 61.04
FinanceReasoning 87.76
GPQA Diamond 81%
Humanity's Last Exam 17.82%
ZebraLogic 96.33%
safety
HealthBench 50.19%
science
PhyBench 47.91
tool-use
BFCL v3 62.01%

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? â–¶

DeepSeek-V3.1-Terminus has the strongest result on Codeforces Elo among the models compared here. See the code-category rows in the table for the full picture.

How many benchmarks are compared? â–¶

This comparison covers 31 benchmarks on which at least one of the selected models has a published score.

Where do the benchmark scores come from? â–¶

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-V3.1-Terminus → AI benchmarks leaderboard → Build a different comparison
Build a different comparison

Models (0 selected)

Benchmarks (auto)

Select at least two models.