Benchmark
HealthBench
Scores are percentages. Higher is better.
Benchmark sourceScores from BenchBoard docs, LLM Stats data.
- Models measured
- 24
- Scores recorded
- 45
- Sources
- 2
- IQ weight
- 2.12
Score distribution
- 56.7%
- Mean
- 58%
- Median
- 7.5%
- Std dev
- 42.5%
- 10th percentile
- 65.4%
- 90th percentile
- 30%
- Min
This benchmark has a weight of 2.12 for chat models, measured over 22 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 5.5adaptive thinking · max effort · default sampling · 5-trial mean · context ≤1M unless noted · raw · no tools · Opus 4.8 grader · max · 5 trials· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 69.4% |
| 2 | Claude Opus 5.5Opus 5.5 System Card · adaptive thinking · max effort unless stated · raw · no tools · Opus 4.8 grader · max · 5 trials· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 68.1% |
| 3 | Claude Opus 5Opus 5.5 System Card · adaptive thinking · max effort unless stated · raw · same Opus 5.5 card setup· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 67.1% |
| 4 | Claude Fable 5.1Opus 5.5 System Card · adaptive thinking · max effort unless stated · raw · same Opus 5.5 card setup· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 66.7% |
| 5 | Claude Sonnet 5.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 65.4% |
| 6 | GPT 5.2GPT-5.2 Thinking · system-card implementation · raw score· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 63.3% |
| 7 | Claude Haiku 5.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 61.6% |
| 8 | Claude Haiku 5.5adaptive thinking · max effort · default sampling · 5-trial mean · context ≤1M unless noted · length-adjusted · max effort · 5-run mean· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 61.6% |
| 9 | Claude Haiku 5.5adaptive thinking · max effort · default sampling · 5-trial mean · context ≤1M unless noted · length-adjusted · xhigh effort · single run· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 61.1% |
| 10 | Claude Opus 5.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 60.6% |
| 11 | Claude Opus 5.5Opus 5.5 System Card · adaptive thinking · max effort unless stated · length-adjusted · no tools · Opus 4.8 grader · max · 5 trials· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 60.6% |
| 12 | Claude Haiku 5.5adaptive thinking · max effort · default sampling · 5-trial mean · context ≤1M unless noted · length-adjusted · high effort · single run· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 60.3% |
| 13 | Claude Haiku 5.5adaptive thinking · max effort · default sampling · 5-trial mean · context ≤1M unless noted · length-adjusted · medium effort · single run· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 60.2% |
| 14 | Qwen3.8 Max· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 60.2% |
| 15 | Qwen3.8 MaxQwen3.8 官方模型卡对照表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 60.2% |
| 16 | Claude Haiku 5.5adaptive thinking · max effort · default sampling · 5-trial mean · context ≤1M unless noted · length-adjusted · low effort · single run· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 59.9% |
| 17 | Claude Fable 5.2· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 59.6% |
| 18 | Claude Opus 4.8Fable/Mythos 5 System Card capability summary · adaptive thinking · max effort · 5 trials unless noted· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 59.3% |
| 19 | Claude Sonnet 5Opus 5.5 System Card · adaptive thinking · max effort unless stated · raw · same Opus 5.5 card setup· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 59.2% |
| 20 | Claude Sonnet 5length-adjusted · Opus 4.8 grader · no tools · 5 trials· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 58.7% |
| 21 | GPT-6.1 Sol· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 58.5% |
| 22 | GPT-6.1 SolGPT-6.1 Sol system card · length-adjusted· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 58.5% |
| 23 | Kimi K2 Thinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 58% |
| 24 | GPT OSS 120Bofficial model card Table 3 · reasoning high · published setting; Table 3 does not further specify tool access · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 57.6% |
| 25 | GPT OSS 120B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 57.6% |
| 26 | GPT-5.6 Sol· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 57% |
| 27 | GPT-5.6 Terra· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 57% |
| 28 | GPT-6.1 SolGPT-6.1 Sol system card · unadjusted· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 56.7% |
| 29 | GPT 5.5Fable/Mythos 5 System Card capability summary · adaptive thinking · max effort · 5 trials unless noted· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 56.5% |
| 30 | GPT 5.5length-adjusted · no tools · GPT-5.5 system-card implementation· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 56.5% |
| 31 | GPT OSS 120Bofficial model card Table 3 · reasoning medium · published setting; Table 3 does not further specify tool access · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 55.9% |
| 32 | GPT-5.6 Luna· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 55.8% |
| 33 | GPT-5.6 SolQwen3.8 官方模型卡对照表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 55.3% |
| 34 | GPT-6.1 Terra· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 55.1% |
| 35 | Qwen3.7 MaxQwen3.8 官方模型卡对照表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 54.5% |
| 36 | GPT-5.3 Chatlatest· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 54.1% |
| 37 | GPT OSS 120Bofficial model card Table 3 · reasoning low · published setting; Table 3 does not further specify tool access · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 53% |
| 38 | Claude Opus 4.8Qwen3.8 官方模型卡对照表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 52.4% |
| 39 | GPT-5.5 Instant· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 51.4% |
| 40 | Claude Sonnet 4.6length-adjusted · Opus 4.8 grader · no tools · Sonnet 5 card rerun· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 49.9% |
| 41 | GPT OSS 20b· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 42.5% |
| 42 | GPT OSS 20bofficial model card Table 3 · reasoning high · published setting; Table 3 does not further specify tool access · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 42.5% |
| 43 | GPT OSS 20bofficial model card Table 3 · reasoning medium · published setting; Table 3 does not further specify tool access · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 41.8% |
| 44 | GPT OSS 20bofficial model card Table 3 · reasoning low · published setting; Table 3 does not further specify tool access · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 40.4% |
| 45 | Seed 2.0 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 30% |
Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links. · Data by LLM Stats