Reasoning benchmark
MLCR-AA leaderboard
Medical Long Context Reasoning (MLCR-AA). Every model the catalog carries a published MLCR-AA value for, ranked by that value.
An open benchmark from Wisedocs, run by Artificial Analysis, measuring how well models reason over long, fragmented medical records with multi-document synthesis.
MLCR-AA ranking
16 models with a published MLCR-AA 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 | Accuracy |
|---|---|---|---|
| 1 |  Claude Fable 5.1 | Anthropic | 71.1 |
| 2 | Claude Fable 5 | Anthropic | 64.4 |
| 3 | Claude Opus 5 | Anthropic | 55.6 |
| 4 | Claude Sonnet 5 | Anthropic | 55.0 |
| 5 |  GLM-5.3-Flash | Z.AI | 51.1 |
| 6 | GLM-5.3 | Z.AI | 48.3 |
| 7 |  Muse Spark 1.3 | Meta | 41.1 |
| 8 |  Kimi K3 | Moonshot AI | 38.3 |
| 9 |  GPT-6 Astra | OpenAI | 35.0 |
| 10 | GPT-5.6 Terra | OpenAI | 31.7 |
| 11 | GPT-5.6 Sol | OpenAI | 26.1 |
| 12 |  Gemini 3.8 Flash | 21.7 | |
| 12 |  Qwen3.8-27B | Alibaba | 21.7 |
| 14 | Muse Glimmer 30B | Meta | 20.0 |
| 15 | GPT-5.6 Luna | OpenAI | 19.4 |
| 16 |  DeepSeek V4 Pro 0813 | DeepSeek | 17.8 |
Evidence key: Observed
Rows are ordered by the value Medical Long Context Reasoning (MLCR-AA) 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.