Capability matrix
Raw benchmark parity rarely decides the pick — tooling, open weights, and provider reliability usually do.
Claude Opus 5 leads on 2 of 2 shared benchmarks; Grok 4.7 is ~73% cheaper per blended million tokens.
| Dimension | Grok 4.7 | Claude Opus 5 |
|---|---|---|
| GPQA Diamond | — | 93.4% |
| HLE | — | 56.3% |
| Terminal-Bench 2.1 | — | 89.1% |
| Terminal-Bench 3.0 | — | 43.3% |
| Terminal-Bench 4.0 | 38% | 51.8% |
| DeepSWE v1.1 | 71% | 74% |
| ProgramBench | — | 37% |
| NL2Repo-Bench | — | 75.3% |
| ExploitGym | — | 22.1% |
| HLE (w/tools) | — | 63.6% |
| Automation-Bench | — | 50.3% |
| Agents' Last Exam | — | 28.6% |
| Chartography (w/tools) | — | 84% |
| BabyVision (w/tools) | — | 94.1% |
| ZeroBench-main (w/tools) | — | 52% |
| Blended price | $4.00 /1M | $15.00 /1M |
| Full priceInput · Output · Cached input | $2.00 · $6.00 · $0.500 | $5.00 · $25.00 · $0.500 |
| Context window | 500K | 1,000K |
Adjust the monthly volume — every model recomputes instantly.
Estimates use list prices and ignore cached input, batch discounts, and off-peak rates.