LiveBench AMPS Hard: leaderboard
Items are machine-generated symbolic maths problems built from the ten hardest AMPS categories with a harder draw of random primitives. The model answers in one turn and its final boxed expression is compared with the ground truth in SymPy; if SymPy cannot confirm a match, the grading script asks a language model to compare the two. This is the AMPS Hard task in LiveBench's maths category, and higher is better.
Metric: Score. Source: livebench.ai. Saturation forecast: Estimated already saturated. 71 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5 (Max) | 99.01 | #12 (Claude Opus 5) |
| 2 | GLM-5.3 | 99 | #39 |
| 3 | Grok 4.5 | 99 | #48 |
| 4 | Gemini 3.8 Flash (High) | 99 | #21 (Gemini 3.8 Flash) |
| 5 | Qwen3.8-Flash-Next | 99 | #72 |
| 6 | Inkling (xHigh) | 99 | #126 (Inkling) |
| 7 | Claude Fable 5.1 (Max) | 99 | #9 (Claude Fable 5.1) |
| 8 | Claude Fable 5 (Max) | 99 | #16 (Claude Fable 5) |
| 9 | Muse Wonder (Max) | 99 | #15 (Muse Wonder) |
| 10 | Muse Spark 1.3 (xHigh) | 99 | #24 (Muse Spark 1.3) |
| 11 | Claude Opus 5.5 (xHigh) | 99 | #3 (Claude Opus 5.5) |
| 12 | DeepSeek V4 Pro | 98 | #102 |
| 13 | DeepSeek V4 Flash | 98 | #137 |
| 14 | GLM-5.2 | 98 | #73 |
| 15 | Qwen 3.7 Max | 98 | #75 |
No result here: #8 Gemini 4 Argon, #10 Muse Wonder Full.
Interactive version: theaggregate.ai/benchmark?slug=livebench-amps-hard · How It Works · Data refreshed daily, snapshot 2026-10-20.