AA Omniscience - Software Engineering (SWE) - Python: leaderboard
Python-language slice of AA Omniscience SWE, testing factual knowledge of Python syntax, libraries, tooling, and engineering practice.
Metric: Accuracy (%). Source: artificialanalysis.ai. Saturation forecast: Around April 2027. 294 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.5 (Thinking) | 74.5 | #83 (Claude Opus 4.5) |
| 2 | Kimi K3 (Max) | 71.5 | #30 (Kimi K3) |
| 3 | Gemini 3 Pro (Preview) (High) | 66.5 | #66 (Gemini 3 Pro (Preview)) |
| 4 | Kimi K3 (Low) | 64.5 | #30 (Kimi K3) |
| 5 | Claude Opus 4.5 (Non-reasoning) | 64 | #83 (Claude Opus 4.5) |
| 6 | Kimi K2.7 Code | 55.5 | #89 |
| 7 | Claude Sonnet 4.5 (Non-reasoning) | 53 | #142 (Claude Sonnet 4.5) |
| 8 | Claude Sonnet 4.5 (Thinking) | 52.5 | #142 (Claude Sonnet 4.5) |
| 9 | Kimi K2.5 (Thinking) | 47 | #144 (Kimi K2.5) |
| 10 | Hy3 | 46.5 | #93 |
| 11 | GPT-5 (High) | 45.5 | #98 (GPT-5) |
| 12 | Inkling (xHigh) | 45 | #126 (Inkling) |
| 13 | DeepSeek V4 Pro (Reasoning, Max Effort) | 45 | #102 (DeepSeek V4 Pro) |
| 14 | GPT-5 (Medium) | 44.5 | #98 (GPT-5) |
| 15 | GPT-5 (Low) | 44 | #98 (GPT-5) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=aa-omniscience-software-engineering-swe-python · How It Works · Data refreshed daily, snapshot 2026-10-20.