Agent Arena Index (Overall)
Arena evaluates models as orchestrators in real, randomly assigned Agent Mode sessions.
The overall index combines task completion, praise versus complaint, steerability and Bash reliability. Scores are expressed in anchored index points, with Bash reliability weighted by one half.
Compare recorded scores and measured cost.
Current recorded best scores, ordered by model release date.
Hover or tap to inspect. Keyboard: ← / → to move, Home / End to jump, Esc to reset.
Claude Fable 5.2: 1448 rating, released Nov 9, 2026
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Max (With Tools + Internet) | 1448 rating | $9.15 | 128K | 1,170 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Thinking High (With Tools + Internet) | 1273 rating | $1.96 | 30.3K | 262 s |
| Max (With Tools + Internet)Best | 1336 rating | $6.00 | 118K | 1,060 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Thinking High (With Tools + Internet) | 1333 rating | $2.60 | 60.7K | 504 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Thinking High (With Tools + Internet) | 1303 rating | $1.35 | 45.1K | 572 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Thinking High (With Tools + Internet) | 1196 rating | $0.76 | 20.3K | 142 s |
| Max (With Tools + Internet)Best | 1296 rating | $3.41 | 146K | 1,010 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Max (With Tools + Internet) | 1291 rating | $3.28 | 35.4K | 707 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Max (With Tools + Internet) | 1282 rating | $4.73 | 31.6K | 633 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Max (With Tools + Internet) | 1272 rating | $7.73 | 61.8K | 723 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Thinking High (With Tools + Internet) | 1271 rating | $1.50 | 63.7K | 317 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Max (With Tools + Internet) | 1241 rating | $0.98 | 37.7K | 473 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Thinking Medium (With Tools + Internet) | 1201 rating | $0.40 | 9.53K | 162 s |
| Max (With Tools + Internet)Best | 1241 rating | $1.03 | 33.6K | 552 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Thinking Medium (With Tools + Internet) | 1155 rating | $0.16 | 7.03K | 103 s |
| Max (With Tools + Internet)Best | 1230 rating | $0.87 | 46.3K | 635 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Thinking High (With Tools + Internet) | 1229 rating | $0.66 | 21.8K | 418 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Extra-High (Tools + Internet) | 1189 rating | $1.41 | 45.5K | 1,010 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Max (With Tools + Internet) | 1178 rating | $1.87 | 40.1K | 322 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Max (With Tools + Internet) | 1176 rating | $4.61 | 60.3K | 599 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Thinking High (With Tools + Internet) | 1173 rating | $3.65 | 24.2K | 273 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Max (With Tools + Internet) | 1145 rating | $1.14 | 32.5K | 764 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Extra-High (Tools + Internet) | 1143 rating | $1.48 | 22.5K | 238 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Thinking High (With Tools + Internet) | 1118 rating | $0.75 | 57.8K | 510 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Extra-High (Tools + Internet) | 1114 rating | $3.35 | 75.9K | 1,080 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Thinking High (With Tools + Internet) | 1110 rating | $0.95 | 26.3K | 225 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Max (With Tools + Internet) | 1110 rating | $0.45 | 28K | 231 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Max (With Tools + Internet) | 1109 rating | $0.14 | 91.6K | 527 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Max (With Tools + Internet) | 1109 rating | $0.02 | 10.1K | 26.9 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Max (With Tools + Internet) | 1107 rating | $1.09 | 22.2K | 199 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Thinking Low (With Tools + Internet) | 1013 rating | $0.0096 | 3.69K | 44.4 s |
| Thinking Medium (With Tools + Internet) | 1056 rating | $0.02 | 9.5K | 96 s |
| Max (With Tools + Internet)Best | 1104 rating | $0.08 | 41.8K | 369 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Thinking High (With Tools + Internet) | 1095 rating | $0.09 | 27.1K | 109 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Max (With Tools + Internet) | 1095 rating | $0.96 | 67.9K | 811 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Max (With Tools + Internet) | 1093 rating | $0.92 | 25.1K | 558 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Max (With Tools + Internet) | 1090 rating | $0.86 | 53.3K | 687 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Extra-High (Tools + Internet) | 1088 rating | $1.35 | 34.4K | 456 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Extra-High (Tools + Internet) | 1061 rating | $0.58 | 17.4K | 180 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Thinking High (With Tools + Internet) | 1057 rating | $0.50 | 38.6K | 336 s |
| Evaluation setting | Score | Cost / task | Output tokens / task | Time / task |
|---|---|---|---|---|
| Extra-High (Tools + Internet) | 979 rating | $0.39 | 25.1K | 104 s |
Source: 06afe0f5-da57-77c9-8000-dccc9d1b0dd0 · All results are from the official leaderboard.
The dataset license covers the test itself; each score source sets its own terms.
Published by Arena · No license recorded; check the publisher’s terms
Unless a license is listed, each source’s own terms apply.
Published by Arena
Source: Arena Agent Arena Index