Model card
Claude 3.5 Sonnet.
AnthropicapiMultimodal LLMProprietary1 current SOTA
Lowest hallucination rate on CC-OCR (0.09%)
§ 02 · Benchmarks
Every benchmark Claude 3.5 Sonnet has a recorded score for.
| # | Benchmark | Area · Task | Metric | Value | Rank | Date | Source |
|---|---|---|---|---|---|---|---|
| 01 | HumanEval | Code & Software Engineering · Code Generation | pass@1 | 92.0% | #11 | 2026-04-20 | source ↗ |
| 02 | MBPP | Code & Software Engineering · Code Generation | pass@1 | 89.2% | #4 | 2026-04-20 | source ↗ |
| 03 | MATH | Language & Knowledge · Mathematical Reasoning | accuracy | 71.1% | #25 | 2026-04-20 | |
| 04 | GSM8K | Language & Knowledge · Mathematical Reasoning | accuracy | 96.4% | #2 | 2026-04-20 | |
| 05 | MMLU | — | accuracy | 88.3% | #9 | 2026-04-20 | |
| 06 | GPQA | — | accuracy | 59.4% | #9 | 2026-04-20 | |
| 07 | CC-OCR | — | multi-scene-f1 | 72.9% | #5 | 2026-04-20 | |
| 08 | CC-OCR | — | kie-f1 | 64.6% | #3 | 2026-04-20 | |
| 09 | SVAMP | — | accuracy | 91.2% | #2 | 2026-04-20 | |
| 10 | ARC-Challenge | — | accuracy | 96.7% | #1 | 2026-04-20 | |
| 11 | CommonsenseQA | — | accuracy | 83.2% | #2 | 2026-04-20 | |
| 12 | WinoGrande | — | accuracy | 85.4% | #2 | 2026-04-20 | |
| 13 | HellaSwag | — | accuracy | 89.0% | #3 | 2026-04-20 | |
| 14 | HotpotQA | — | f1 | 68.5% | #2 | 2026-04-20 | |
| 15 | LogiQA | Language & Knowledge · Logical Reasoning | accuracy | 53.8% | #2 | 2026-04-20 | |
| 16 | ReClor | Language & Knowledge · Logical Reasoning | accuracy | 68.9% | #2 | 2026-04-20 | |
| 17 | StrategyQA | — | accuracy | 79.8% | #2 | 2026-04-20 | |
| 18 | MAWPS | — | accuracy | 95.8% | #2 | 2026-04-20 |
Rank column shows this model’s position vs all other models scored on the same benchmark + metric (competitors after the slash). #1 in red means current SOTA. Sorted by rank, then newest result.
§ 03 · Strengths by area
Where Claude 3.5 Sonnet actually performs.
§ 06 · Sources & freshness
Where these numbers come from.
src
16
results
codesota-api
2
results
0 of 18 rows marked verified.