Harvey's Legal Agent BenchmarkBrowse 284 families
Harvey's Legal Agent Benchmark
Can the model produce useful work in a professional knowledge-work setting?
- Domain
- Knowledge Work
- Ranked models
- 7
- Reported values
- 7
- Reports
- 2 · lab
- Direction
- Higher is better
- Status
- Version not specified
Rankings
Best reported score per model across effort settings
| # | Model | Best setting | Score |
|---|---|---|---|
| 1 | Grok 4.7SpaceXAI | xhighcost n/a1 value · Sep 21 · source | 19.6% |
| 2 | Gemini 3.8 FlashGoogle | reportedcost n/a1 value · Sep 2 · source | 10.0% |
| 3 | Gemini 3.7 FlashGoogle | reportedcost n/a1 value · Sep 2 · source | 8.8% |
| 4 | Claude Opus 5Anthropic | reportedcost n/a1 value · Sep 2 · source | 6.7% |
| 5 | Claude Sonnet 5Anthropic | reportedcost n/a1 value · Sep 2 · source | 5.0% |
| 6 | GPT-5.6 SolOpenAI | reportedcost n/a1 value · Sep 2 · source | 2.5% |
| 7 | GPT-5.6 TerraOpenAI | reportedcost n/a1 value · Sep 2 · source | 0.8% |
Effort curve
Score vs. reported cost. Lines connect complete sweeps only.
Definition and comparison boundarypublic methodology
Can the model produce useful work in a professional knowledge-work setting?
Professional analysis, document, finance, office, or domain-specific tasks. Higher is better. The value is the percentage reported in this lab's table.
Harness, tools, prompts, attempt count, and benchmark version can materially change the result. Compare within one reporting context.
SpaceXAI reports 19.6% for Grok 4.7 at xhigh; the table labels this metric as legal work but does not print the harness or task revision. Complex legal workflows; the model card labels this metric All pass rate.
Method / source ↗