Legacy-Bench: leaderboard
Coding agents fix bugs, add features and migrate code in COBOL, Java 7, BASIC, C89, Fortran and Assembly, each task checked by hidden tests; the score is the pass rate over 3 Fireworks runs of 100 tasks. SWE-bench-style boards use modern Python and JavaScript repositories instead.
Metric: Tasks passed, mean of 3 runs of 100 tasks (%). Source: fireworks.ai. Saturation forecast: Around 2031. 13 models tracked. Updated 16 Nov 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Kimi K3.1 Pro | Max | 63 | #29 |
| 2 | Muse Spark 1.3 | Max | 62 | #43 |
| 3 | Claude Opus 5.5 | Medium | 62 | #5 |
| 4 | DeepSeek V4.1 Flash | Max | 57.67 | #62 |
| 5 | DeepSeek V4.1 Flash (1017) | Max | 56.33 | #41 |
| 6 | GPT-6 Astra | Max | 55.67 | #6 |
| 7 | GPT-6.1 Sol | Max | 55 | #10 |
| 8 | Grok 4.7 | Max | 54 | #47 |
| 9 | Kimi K3 | Max | 52.67 | #56 |
| 10 | Gemini 3.8 Flash | 52.33 | #36 | |
| 11 | GLM-5.3 | Max | 50.67 | #68 |
| 12 | Muse Spark Open | Max | 50 | #81 |
| 13 | Gemini 4 Argon | High | 44.33 | #12 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon.
Interactive version: theaggregate.ai/benchmark?slug=legacy-bench · How It Works · Data refreshed daily, snapshot 2026-11-16.