Each card below is one named authority's own current view of Claude Opus 4.7. Model Gauntlet mirrors these sources with attribution and does not combine them into a single score or rank Claude Opus 4.7 across them.
LMArena overall
Human preference, style-controlled text
- LMArena rating
- 1493.7
- Board rank
- #6
- Reported interval
- 1489.5 to 1497.9
- Battles
- 50,890
Published July 20, 2026. LMArena leaderboard dataset, CC-BY-4.0.
Overall leaderboard →LMArena coding arena
Human preference, coding arena
- LMArena rating
- 1548.7
- Board rank
- #3
- Reported interval
- 1542.2 to 1555.2
- Battles
- 14,491
Published July 20, 2026. LMArena leaderboard dataset, CC-BY-4.0.
Coding arena board →LMArena creative writing arena
Human preference, creative writing arena
- LMArena rating
- 1482.6
- Board rank
- #5
- Reported interval
- 1474.7 to 1490.4
- Battles
- 8,647
Published July 20, 2026. LMArena leaderboard dataset, CC-BY-4.0.
Creative writing arena board →LMArena instruction following arena
Human preference, instruction following arena
- LMArena rating
- 1492.9
- Board rank
- #6
- Reported interval
- 1486.9 to 1499.0
- Battles
- 17,352
Published July 20, 2026. LMArena leaderboard dataset, CC-BY-4.0.
Instruction following arena board →LMArena hard prompts arena
Human preference, hard prompts arena
- LMArena rating
- 1519.1
- Board rank
- #5
- Reported interval
- 1514.1 to 1524.2
- Battles
- 33,785
Published July 20, 2026. LMArena leaderboard dataset, CC-BY-4.0.
Hard prompts arena board →LMArena multi-turn arena
Human preference, multi-turn arena
- LMArena rating
- 1518.8
- Board rank
- #1
- Reported interval
- 1511.1 to 1526.4
- Battles
- 9,076
Published July 20, 2026. LMArena leaderboard dataset, CC-BY-4.0.
Multi-turn arena board →LMArena long queries arena
Human preference, long queries arena
- LMArena rating
- 1508.8
- Board rank
- #5
- Reported interval
- 1502.9 to 1514.7
- Battles
- 22,653
Published July 20, 2026. LMArena leaderboard dataset, CC-BY-4.0.
Long queries arena board →LMArena math arena
Human preference, math arena
- LMArena rating
- 1493.4
- Board rank
- #10
- Reported interval
- 1481.2 to 1505.6
- Battles
- 2,627
Published July 20, 2026. LMArena leaderboard dataset, CC-BY-4.0.
Math arena board →LMArena vision arena
Human preference, vision arena
- LMArena rating
- 1315.5
- Board rank
- #4
- Reported interval
- 1308.5 to 1322.5
- Battles
- 19,049
Published July 12, 2026. LMArena leaderboard dataset, CC-BY-4.0.
Vision arena board →LMArena web development arena
Human preference, web development arena
- LMArena rating
- 1561.9
- Board rank
- #2
- Reported interval
- 1549.8 to 1574.0
- Battles
- 2,999
Published May 4, 2026. LMArena leaderboard dataset, CC-BY-4.0.
Web development arena board →LMArena search arena
Human preference, search arena
- LMArena rating
- 1234.9
- Board rank
- #3
- Reported interval
- 1229.3 to 1240.4
- Battles
- 62,037
Published July 12, 2026. LMArena leaderboard dataset, CC-BY-4.0.
Search arena board →LMArena document arena
Human preference, document arena
- LMArena rating
- 1501.2
- Board rank
- #5
- Reported interval
- 1494.7 to 1507.7
- Battles
- 18,185
Published July 12, 2026. LMArena leaderboard dataset, CC-BY-4.0.
Document arena board →EQ-Bench 3
Emotional intelligence
- Judge Elo
- 1460.9
- Source rank
- #1
- Source label
- claude-opus-4-7
EQ-Bench 3 is a single-source subjective judge-based pairwise Elo for emotional-intelligence roleplays, appropriate only for that lane.
Source last updated May 10, 2026. EQ-Bench 3 canonical Elo leaderboard, EQ-bench/eqbench3, MIT.
EQ-Bench 3 source ↗