BlogLLM Benchmark Scores as Published, Side by Side
DynaMath: Results as Published
Every result the catalog holds for DynaMath: 14 models from 5 developers, read from 3 pages. Each number is as printed by the company named beside it; nothing is ranked or rescored. Models are listed by developer, newest first.
Area Multimodal UnderstandingMaintainer Chengke Zou and co-authors (University of Illinois Urbana-Champaign and University of California, Berkeley)Introduced 2024Data version 2026103001, as of 2026-10-30
DynaMath tests whether visual math reasoning still holds after a problem is rewritten. It has 501 seed questions, each stored as a Python program that can change numbers, graphs, or figures. The authors generate 10 variants of each seed, which makes 5,010 concrete questions. Average-case accuracy is the share of concrete questions answered correctly, and worst-case accuracy is the share of seeds for which every variant is correct.
DynaMath
| Model | Value | Printed by | Setup as printed |
|---|---|---|---|
| GPT-5.2OpenAI, released 2025-12-11 | 86.8% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| GPT-5 miniOpenAI, released 2025-08-07 | 81.4% | Alibaba, Qwen3.5-122B-A10B model cardchecked 2026-10-15 | - |
| Claude Opus 4.5Anthropic, released 2025-11-24 | 79.7% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15; printed on 1 more page | - |
| Claude Sonnet 4.5Anthropic, released 2025-09-29 | 78.8% | Alibaba, Qwen3.5-122B-A10B model cardchecked 2026-10-15 | - |
| Gemma 4 31BGoogle, released 2026-04-02 | 79.5% | Alibaba, Qwen3.6-27B model cardchecked 2026-10-15 | - |
| Gemini 3 ProGoogle, released 2025-11-18 | 85.1% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| Qwen3.6-27BAlibaba, released 2026-04-22 | 85.6% | Alibaba, Qwen3.6-27B model cardchecked 2026-10-15 | - |
| Qwen3.6-35B-A3BAlibaba, released 2026-04-16 | 82.8% | Alibaba, Qwen3.6-27B model cardchecked 2026-10-15 | - |
| Qwen3.5-122B-A10BAlibaba, released 2026-02-24 | 85.9% | Alibaba, Qwen3.5-122B-A10B model cardchecked 2026-10-15 | - |
| Qwen3.5-27BAlibaba, released 2026-02-24 | 87.7% | Alibaba, Qwen3.5-122B-A10B model cardchecked 2026-10-15; printed on 1 more page | - |
| Qwen3.5-35B-A3BAlibaba, released 2026-02-24 | 85% | Alibaba, Qwen3.5-122B-A10B model cardchecked 2026-10-15 | - |
| Qwen3.5-397B-A17BAlibaba, released 2026-02-16 | 86.3% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15; printed on 1 more page | - |
| Qwen3-VL-235B-A22BAlibaba, released 2025-09-23 | 82.8% | Alibaba, Qwen3.5-122B-A10B model cardchecked 2026-10-15; printed on 1 more page | - |
| Kimi K2.5Moonshot AI, released 2026-01-27 | 84.4% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
Every number here is read from the same catalog file as the benchmark table, data version 2026103001, and is the number the table shows for that model and test. Where one page printed a model under several setups, each value is listed. The catalog is public in armanobosyan/sugra-llm-benchmarks; a wrong number can be reported from the table, with Report an error under the value, or as an issue in that repository.