Loading page…
Preferred languages
More languages supported by Google Translate
Open the selector to load the current language catalog.
Translations are provided by Google Translate.
Loading page…
Preferred languages
More languages supported by Google Translate
Open the selector to load the current language catalog.
Translations are provided by Google Translate.
Knowledge benchmark
Every model the catalog carries a published MMLU-Redux value for, ranked by that value.
A harder refresh of MMLU intended to keep broad knowledge evaluation useful after the original benchmark became too easy for frontier models. 19 models in the ModelChoice catalog have a published value.
Each point is a model with a published MMLU-Redux value and complete token prices, on the 75% input / 25% output blend. Frontier models are spaced evenly along the cost axis.
19 models with a published MMLU-Redux value, ordered by that value, highest first. Models the source has not scored on this benchmark are not listed — they are unmeasured, not last.
| Rank | Model | Provider | Multiple choice questions |
|---|---|---|---|
| 1 | Anthropic | 96.6 | |
| 2 | Alibaba | 95 | |
| 3 | Alibaba | 94.9 |
What the benchmark measures and how its value is published, from the source’s own record.
MMLU-Redux is a knowledge benchmark published by Qwen3.6 launch benchmarks. A harder refresh of MMLU intended to keep broad knowledge evaluation useful after the original benchmark became too easy for frontier models. Its evaluation set is described by the source as broad academic QA. The source rates its difficulty as advanced general knowledge.
Each model's result is published by Qwen3.6 launch benchmarks as multiple choice questions. ModelChoice prints that value exactly as published, and never converts or rescales it. This page ranks the 19 models that carry a published value, highest value first; a model the source never scored is left out rather than ranked at zero. The source does not state whether a higher value is the better result, so neither does this page.
| 4 | Moonshot AI | 94.5 |
| 4 | Alibaba | 94.5 |
| 4 | Alibaba | 94.5 |
| 7 | Alibaba | 93.5 |
| 8 | DeepSeek | 93.4 |
| 9 | Alibaba | 92.7 |
| 10 | Alibaba | 90.9 |
| 11 | Alibaba | 90 |
| 12 | Alibaba | 89.2 |
| 13 | PMTernary Bonsai 2 27B | Prism ML | 89.09 |
| 14 | JEMellum2.1-12B-A2.5B-Thinking | JetBrains | 87.8 |
| 15 | JEMellum2-12B-A2.5B-Thinking | JetBrains | 86.2 |
| 16 | Alibaba | 85.7 |
| 17 | OPMiniCPM5-2B | OpenBMB | 84.7 |
| 18 | JEMellum2-12B-A2.5B-Instruct | JetBrains | 78.1 |
| 19 | OPMiniCPM5-1B | OpenBMB | 70.06 |
Evidence key: Observed
Rows are ordered by the value Qwen3.6 launch benchmarks published, highest first. The source does not state whether a higher value is the better result, so this page does not either: for a benchmark that measures a rate of failure, read the table from the bottom.