AnthropicAuto-detectedClaude 3 Haiku
Best configuration ranks #570 of 610 on the BenchLeader Index at 36.9 ±4.3. Last measured 2 Sept 2026. Released 7 Mar 2024.
- Blended price
- $0.500/M
- $0.250 in · $1.25 out
- Output speed
- 80 tok/s
- First answer
- 0.46 s
- Context
- 200k
- Overall index37
- Reasoning33
- Coding32
- Agents & tools34
- Maths27
- Knowledge41
- Instruction following39
- Human preference42
- Multimodal20
- Long context39
- Composite36
Benchmark results
One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | default | Source |
|---|---|---|
| GPQA Diamond | 36.3%#238 | Epoch AI Benchmarking Hub |
| LMArena Hard Prompts | 1265#260 | LMArena |
| GPQA Diamond (AA)not in index | 37.4%#461 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 4.1%#460 | Artificial Analysis |
Coding
| Benchmark | default | Source |
|---|---|---|
| WeirdML | 9.8%#145 | WeirdML |
| LMArena Coding | 1301#254 | LMArena |
| SWE-bench Verified (any scaffold)not in index | 40.6%#49 | SWE-bench |
Agents & tools
| Benchmark | default | Source |
|---|---|---|
| Terminal-Bench Hard | 0.8%#332 | Artificial Analysis |
| τ²-Bench Telecom (AA)not in index | 21.1%#312 | Artificial Analysis |
Maths
| Benchmark | default | Source |
|---|---|---|
| OTIS Mock AIME | 1.8%#243 | Epoch AI Benchmarking Hub |
| MATH Level 5 | 14.9%#78 | Epoch AI Benchmarking Hub |
Knowledge
| Benchmark | default | Source |
|---|---|---|
| AA-Omniscience | -48.6#307 | Artificial Analysis |
Instruction following
| Benchmark | default | Source |
|---|---|---|
| IFBench | 36.1%#292 | Artificial Analysis |
Human preference
| Benchmark | default | Source |
|---|---|---|
| LMArena Text | 1261#261 | LMArena |
Multimodal
| Benchmark | default | Source |
|---|---|---|
| LMArena Vision | 950#133 | LMArena |
| MMMU-Pro | 30.8%#238 | Artificial Analysis |
| MMMU (validation) | 50.2%#41 | MMMU |
Long context
| Benchmark | default | Source |
|---|---|---|
| AA-LCR | 27.7%#332 | Artificial Analysis |
Composite
| Benchmark | default | Source |
|---|---|---|
| Epoch Capabilities Indexnot in index | 118.3#171 | Epoch AI Benchmarking Hub |
| AA Intelligence Index | 5.6#483 | Artificial Analysis |
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| Amazon Bedrock | 80 tok/s | 0.46 s | $0.250 | $1.25 | 200k | – |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0005 | – | 4.2 s |
| Summarise a 30-page report | 12,000 / 600 | $0.0038 | – | 8.0 s |
| Code edit | 6,000 / 1,500 | $0.0034 | – | 19.2 s |
| Agentic coding session | 60,000 / 4,000 | $0.020 | – | 50.5 s |
| Structured extraction | 2,000 / 200 | $0.0008 | – | 3.0 s |
See also
Data as of 9 Sept 2026. Compare with another model.