ProgramBench Almost: leaderboard
Companion ProgramBench metric that counts near-complete program recreations: tasks where the generated implementation passes most hidden behavioral tests but does not fully resolve the benchmark task.
Metric: Almost (%). Source: programbench.com. Saturation forecast: Around September 2027. 25 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5 (xHigh) | 37 | #12 (Claude Opus 5) |
| 2 | Muse Spark 1.3 (Max) | 25 | #24 (Muse Spark 1.3) |
| 3 | Muse Spark 1.3 (xHigh) | 16.5 | #24 (Muse Spark 1.3) |
| 4 | Claude Opus 4.8 (xHigh) | 16.5 | #37 (Claude Opus 4.8) |
| 5 | GPT-5.6 Sol (xHigh) | 15.5 | #19 (GPT-5.6 Sol) |
| 6 | GPT-5.5 (xHigh) | 13.5 | #31 (GPT-5.5) |
| 7 | GLM-5.2 | 8.5 | #73 |
| 8 | Gemini 3.7 Flash | 5.5 | #23 |
| 9 | GPT-5.5 (High) | 5 | #31 (GPT-5.5) |
| 10 | Muse Spark 1.2 (xHigh) | 4.5 | #62 (Muse Spark 1.2) |
| 11 | Claude Opus 4.7 (xHigh) | 4.5 | #49 (Claude Opus 4.7) |
| 12 | Gemini 3.6 Flash | 4 | #51 |
| 13 | Muse Spark 1.1 (xHigh) | 4 | #52 (Muse Spark 1.1) |
| 14 | Claude Opus 4.7 | 3 | #49 |
| 15 | Gemini 3.5 Flash | 3 | #60 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=programbench-almost · How It Works · Data refreshed daily, snapshot 2026-10-20.