AA Omniscience - Software Engineering (SWE) - Java: leaderboard
Java-language slice of AA Omniscience SWE, testing factual knowledge of Java syntax, libraries, tooling, and engineering practice.
Metric: Accuracy (%). Source: artificialanalysis.ai. Saturation forecast: Around September 2028. 294 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro (Preview) (High) | 55 | #66 (Gemini 3 Pro (Preview)) |
| 2 | Claude Opus 4.5 (Thinking) | 53 | #83 (Claude Opus 4.5) |
| 3 | Kimi K3 (Max) | 52 | #30 (Kimi K3) |
| 4 | Kimi K3 (Low) | 48 | #30 (Kimi K3) |
| 5 | Claude Opus 4.5 (Non-reasoning) | 45 | #83 (Claude Opus 4.5) |
| 6 | Kimi K2.7 Code | 43 | #89 |
| 7 | GPT-5 Codex (High) | 42 | #127 (GPT-5 Codex) |
| 8 | DeepSeek V4 Pro (Reasoning, Max Effort) | 39.39 | #102 (DeepSeek V4 Pro) |
| 9 | Grok 4 | 38 | #181 |
| 10 | Hy3 | 37 | #93 |
| 11 | GPT-5 (Low) | 37 | #98 (GPT-5) |
| 12 | GPT-5 (High) | 36 | #98 (GPT-5) |
| 13 | DeepSeek V4 Pro (Non-reasoning) | 36 | #102 (DeepSeek V4 Pro) |
| 14 | Kimi K2.5 (Thinking) | 35 | #144 (Kimi K2.5) |
| 15 | Kimi K2.6 | 34 | #104 |
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-java · How It Works · Data refreshed daily, snapshot 2026-10-20.