What is the SciCode leaderboard?

The SciCode leaderboard ranks 12 AI models based on their performance on this benchmark. Currently, Gemini 3.1 Pro by Google leads with a score of 0.590. The average score across all models is 0.438.

What is the highest SciCode score?

The highest SciCode score is 0.590, achieved by Gemini 3.1 Pro from Google.

How many models are evaluated on SciCode?

12 models have been evaluated on the SciCode benchmark, with 0 verified results and 12 self-reported results.

Where can I find the SciCode paper?

The SciCode paper is available at https://arxiv.org/abs/2407.13168. The paper details the methodology, dataset construction, and evaluation criteria.

What categories does SciCode cover?

SciCode is categorized under math, physics, reasoning, biology, chemistry, and code. The benchmark evaluates text models.

What is the best open-source model on SciCode?

Kimi K2.6 by Moonshot AI is the top-ranked open-source model on SciCode, with a score of 0.522 (rank #3).

Which model offers the best value on SciCode?

Among models scoring within 10% of the leader, Qwen3.7 Max from Alibaba Cloud / Qwen Team is the cheapest, at $1.25 per million input tokens with a score of 0.535.

How recent are the SciCode leaderboard results?

The SciCode leaderboard was last updated in June 2026 and currently includes 12 evaluated models.

All benchmarks

SciCode

SciCode is a research coding benchmark curated by scientists that challenges language models to code solutions for scientific problems. It contains 338 subproblems decomposed from 80 challenging main problems across 16 natural science sub-fields including mathematics, physics, chemistry, biology, and materials science. Problems require knowledge recall, reasoning, and code synthesis skills.

Gemini 3.1 Pro from Google currently leads the SciCode leaderboard with a score of 0.590 across 12 evaluated AI models.

Paper