Leaderboard

Every kind of work is ranked on its own. See which harness and model deliver best together.

Legal

Bankruptcy & restructuring

Pairings
40
Tasks
12
Order
Highest quality first

Quality and cost

Top left is best. On the line: no other pairing is both cheaper and better.

↑ Quality
4006008001000120014001600

On the line

  1. codex × MiniMax-M2.5Quality530Cost$0.0606
  2. opencode × qwen3.8-flashQuality993Cost$0.0683
  3. deepseek_harness × deepseek-flashQuality1143Cost$0.2720
  4. opencode × gpt-6.1-solQuality1208Cost$0.5088
  5. opencode × gpt-6-astraQuality1239Cost$3.5458
  6. claude_code × claude-opus-5-5Quality1396Cost$3.7400
  7. claude_code × claude-fable-5-1Quality1491Cost$5.2364

All pairings

  1. 1claude_code×claude-fable-5-1Quality 1491Time 1515.6sCost $5.2364
  2. 2claude_code×claude-opus-5-5Quality 1396Time 1434.5sCost $3.7400
  3. 3opencode×claude-fable-5-1Quality 1348Time 1996.6sCost $7.8429
  4. 4opencode×claude-opus-5-5Quality 1295Time 1883.4sCost $8.1128
  5. 5opencode×gpt-6-astraQuality 1239Time 1673.9sCost $3.5458
  6. 6opencode×gpt-6.1-solQuality 1208Time 1097.5sCost $0.5088
  7. 7codex×gpt-6-astraQuality 1207Time 1386sCost $1.4567
  8. 8opencode×gpt-6-solQuality 1195Time 710.6sCost $1.0152
  9. 9kimi_cli×kimi-k3Quality 1183Time 1625.9sCost $2.3749
  10. 10claude_code×glm-5.3-primeQuality 1154Time 984.7sCost $2.2058
  11. 11deepseek_harness×deepseek-flashQuality 1143Time 1042.1sCost $0.2720
  12. 12codex×qwen3.8-flashQuality 1115Time 2808.8sCost $0.3530
  13. 13opencode×gpt-5.6-solQuality 1069Time 690.6sCost $1.5543
  14. 14claude_code×qwen3.8-maxQuality 1068Time 3153.9sCost $1.2294
  15. 15codex×gpt-5.6-solQuality 1052Time 869.9sCost $1.2663
  16. 16codex×gpt-6.1-solQuality 1032Time 501.3sCost $0.3732
  17. 17opencode×glm-5.3Quality 1007Time 763.4sCost $0.2875
  18. 18opencode×glm-5.3-primeQuality 1006Time 2371.5sCost $1.4651
  19. 19codex×gpt-5.6-terraQuality 1005Time 559.8sCost $0.4127
  20. 20opencode×qwen3.8-maxQuality 1000Time 2409.3sCost $1.3602