Anthropic's Feb-2026 flagship. Strong agentic coding (80.84% SWE-bench Verified) and reasoning (91.3% GPQA Diamond). $5/$25 per Mtok, full 1M-token context.
Key specs
| Type | API (hosted) |
|---|
| Parameters | — |
|---|
| Architecture | undisclosed |
|---|
| Context window | 1000K tokens |
|---|
| Knowledge cutoff | — |
|---|
| Modalities | text, image |
|---|
| Recommended backends | — |
|---|
| Minimum viable rig | — |
|---|
Benchmark scores
| GPQA Diamond | 91.31% |
|---|
| SWE-bench Verified | 80.84% |
|---|
| AIME | — |
|---|
| MMLU-Pro | — |
|---|
| BFCL v3 (tool use) | — |
|---|
| Composite score | 8.6 |
|---|
| Community rating | No reviews yet |
|---|
API pricing (per 1M tokens)
| Provider | Input | Output | Free tier |
|---|
| Anthropic | $5 | $25 | No |
| AWS Bedrock | $5 | $25 | No |