AA Omniscience - Software Engineering (SWE) - Julia: leaderboard

Metric: Accuracy (%). Source: artificialanalysis.ai. 294 models tracked.

Top models

#ModelScore
1Claude Opus 4.5 (Thinking)68
2Gemini 3 Pro (Preview) (High)56
3Claude Opus 4.5 (Non-reasoning)56
4Claude Sonnet 4.5 (Non-reasoning)52
5DeepSeek V4 Pro (0813) (Reasoning, Max Effort)52
6Kimi K2.7 Code48
7Kimi K3 (Max)48
8Kimi K3 (Low)48
9GPT-5 (High)44
10GPT-5 (Medium)40
11Gemini 2.5 Pro36
12Kimi K2.636
13Claude Sonnet 4.5 (Thinking)36
14O336
15Qwen 3 235B A22B 2507 (Thinking)36

Interactive version: theaggregate.ai/benchmark?slug=aa-omniscience-software-engineering-swe-julia · How It Works · Data refreshed daily, snapshot 2026-09-05.