LMArena Document
Arena human-vote leaderboard: rating with confidence interval, vote count and rank. One row per model and category.
No filters. Showing 46 of 46 rows.
1[1–4] AnthropicDossier | max | No harness | Fable 5.1, Arena rating: 1,515 (scale 1350.0 to 1514.8), 95% confidence interval 1499.7 to 1530.01,515(1499.7–1530.0) | 1453 seeds | Cost kind unknown |
2[1–4] AnthropicDossierTied #1 | high | No harness | Opus 5, Arena rating: 1,509 (scale 1350.0 to 1514.8), 95% confidence interval 1503.0 to 1514.31,509(1503.0–1514.3) | 16189 seeds | Cost kind unknown |
3[1–4] AnthropicDossierTied #1 | high | No harness | Opus 4.6, Arena rating: 1,508 (scale 1350.0 to 1514.8), 95% confidence interval 1501.2 to 1514.01,508(1501.2–1514.0) | 29456 seeds | Cost kind unknown |
4[1–4] AnthropicDossierTied #1 | No effort set | No harness | Opus 4.6, Arena rating: 1,506 (scale 1350.0 to 1514.8), 95% confidence interval 1501.0 to 1511.41,506(1501.0–1511.4) | 47264 seeds | Cost kind unknown |
5[5–7] AnthropicDossier | high | No harness | Fable 5, Arena rating: 1,497 (scale 1350.0 to 1514.8), 95% confidence interval 1490.0 to 1503.31,497(1490.0–1503.3) | 12357 seeds | Cost kind unknown |
6[5–7] AnthropicDossier | high | No harness | Opus 4.7, Arena rating: 1,496 (scale 1350.0 to 1514.8), 95% confidence interval 1489.3 to 1502.11,496(1489.3–1502.1) | 23327 seeds | Cost kind unknown |
7[5–7] AnthropicDossier | No effort set | No harness | Opus 4.7, Arena rating: 1,495 (scale 1350.0 to 1514.8), 95% confidence interval 1488.4 to 1501.21,495(1488.4–1501.2) | 23628 seeds | Cost kind unknown |
8[8–12] OpenAIDossier | high | No harness | GPT-5.5, Arena rating: 1,484 (scale 1350.0 to 1514.8), 95% confidence interval 1478.9 to 1489.61,484(1478.9–1489.6) | 27324 seeds | Cost kind unknown |
9[8–15] GoogleDossier | high | No harness | Gemini 3.8 Flash, Arena rating: 1,482 (scale 1350.0 to 1514.8), 95% confidence interval 1474.0 to 1489.91,482(1474.0–1489.9) | 6950 seeds | Cost kind unknown |
10[8–13] AnthropicDossier | No effort set | No harness | Sonnet 4.6, Arena rating: 1,482 (scale 1350.0 to 1514.8), 95% confidence interval 1476.0 to 1487.71,482(1476.0–1487.7) | 61055 seeds | Cost kind unknown |
11[8–13] OpenAIDossier | No effort set | No harness | GPT-5.5, Arena rating: 1,481 (scale 1350.0 to 1514.8), 95% confidence interval 1475.6 to 1486.21,481(1475.6–1486.2) | 27751 seeds | Cost kind unknown |
12[8–15] OpenAIDossier | xhigh | No harness | GPT-5.6 Sol, Arena rating: 1,480 (scale 1350.0 to 1514.8), 95% confidence interval 1473.2 to 1486.81,480(1473.2–1486.8) | 9649 seeds | Cost kind unknown |
13[9–17] GoogleDossier | high | No harness | Gemini 3.7 Flash, Arena rating: 1,476 (scale 1350.0 to 1514.8), 95% confidence interval 1468.9 to 1484.11,476(1468.9–1484.1) | 7680 seeds | Cost kind unknown |
14[11–18] AnthropicDossier | high | No harness | Opus 4.8, Arena rating: 1,475 (scale 1350.0 to 1514.8), 95% confidence interval 1467.7 to 1481.41,475(1467.7–1481.4) | 12069 seeds | Cost kind unknown |
15[12–19] OpenAIDossier | No effort set | No harness | GPT-5.4, Arena rating: 1,473 (scale 1350.0 to 1514.8), 95% confidence interval 1467.0 to 1478.91,473(1467.0–1478.9) | 35183 seeds | Cost kind unknown |
16[11–21] MetaDossier | max | No harness | Muse Spark 1.3, Arena rating: 1,473 (scale 1350.0 to 1514.8), 95% confidence interval 1465.1 to 1480.71,473(1465.1–1480.7) | 7205 seeds | Cost kind unknown |
17[15–23] OpenAIDossier | xhigh | No harness | GPT-5.6 Terra, Arena rating: 1,468 (scale 1350.0 to 1514.8), 95% confidence interval 1461.2 to 1474.11,468(1461.2–1474.1) | 11412 seeds | Cost kind unknown |
18[13–24] OpenAIDossier | max | No harness | GPT-6 Astra, Arena rating: 1,467 (scale 1350.0 to 1514.8), 95% confidence interval 1458.7 to 1475.81,467(1458.7–1475.8) | 5723 seeds | Cost kind unknown |
19[14–25] AnthropicDossier | No effort set | No harness | Opus 4.5, Arena rating: 1,466 (scale 1350.0 to 1514.8), 95% confidence interval 1455.5 to 1475.91,466(1455.5–1475.9) | 8304 seeds | Cost kind unknown |
20[16–24] AnthropicDossier | No effort set | No harness | Opus 4.8, Arena rating: 1,465 (scale 1350.0 to 1514.8), 95% confidence interval 1458.7 to 1471.41,465(1458.7–1471.4) | 14729 seeds | Cost kind unknown |
21[17–25] MetaDossier | No effort set | No harness | Muse Spark 1.1, Arena rating: 1,463 (scale 1350.0 to 1514.8), 95% confidence interval 1456.4 to 1470.61,463(1456.4–1470.6) | 8934 seeds | Cost kind unknown |
22[17–27] GoogleDossier | high | No harness | Gemini 3.5 Flash, Arena rating: 1,462 (scale 1350.0 to 1514.8), 95% confidence interval 1452.6 to 1470.91,462(1452.6–1470.9) | 4248 seeds | Cost kind unknown |
23[17–27] GoogleDossier | medium | No harness | Gemini 3.5 Flash, Arena rating: 1,461 (scale 1350.0 to 1514.8), 95% confidence interval 1452.9 to 1468.41,461(1452.9–1468.4) | 6835 seeds | Cost kind unknown |
24[19–26] AnthropicDossier | high | No harness | Sonnet 5, Arena rating: 1,460 (scale 1350.0 to 1514.8), 95% confidence interval 1453.0 to 1466.11,460(1453.0–1466.1) | 11044 seeds | Cost kind unknown |
25[19–29] GoogleDossier | high | No harness | Gemini 3.6 Flash, Arena rating: 1,456 (scale 1350.0 to 1514.8), 95% confidence interval 1445.4 to 1466.81,456(1445.4–1466.8) | 3031 seeds | Cost kind unknown |
26[22–30] OpenAIDossier | xhigh | No harness | GPT-5.6 Luna, Arena rating: 1,454 (scale 1350.0 to 1514.8), 95% confidence interval 1446.8 to 1460.81,454(1446.8–1460.8) | 8832 seeds | Cost kind unknown |
27[24–30] xAIDossier | No effort set | No harness | Grok 4.5, Arena rating: 1,451 (scale 1350.0 to 1514.8), 95% confidence interval 1443.9 to 1459.11,451(1443.9–1459.1) | 7447 seeds | Cost kind unknown |
28[24–31] Moonshot AIDossier | No effort set | No harness | Kimi K2.6, Arena rating: 1,450 (scale 1350.0 to 1514.8), 95% confidence interval 1442.9 to 1458.01,450(1442.9–1458.0) | 11809 seeds | Cost kind unknown |
29[25–31] AnthropicDossier | No effort set | No harness | Sonnet 4.5, Arena rating: 1,449 (scale 1350.0 to 1514.8), 95% confidence interval 1443.3 to 1455.41,449(1443.3–1455.4) | 32916 seeds | Cost kind unknown |
30[23–38] MetaDossier | No effort set | No harness | Muse Spark, Arena rating: 1,444 (scale 1350.0 to 1514.8), 95% confidence interval 1425.4 to 1461.71,444(1425.4–1461.7) | 1116 seeds | Cost kind unknown |
31[29–33] GoogleDossier | No effort set | No harness | Gemini 3.1 Pro Preview, Arena rating: 1,443 (scale 1350.0 to 1514.8), 95% confidence interval 1438.1 to 1448.41,443(1438.1–1448.4) | 52244 seeds | Cost kind unknown |
32[30–37] xAIDossier | high | No harness | Grok 4.6, Arena rating: 1,438 (scale 1350.0 to 1514.8), 95% confidence interval 1430.3 to 1445.81,438(1430.3–1445.8) | 7669 seeds | Cost kind unknown |
33[31–36] AlibabaDossier | No effort set | No harness | Qwen3.7 Plus, Arena rating: 1,437 (scale 1350.0 to 1514.8), 95% confidence interval 1430.0 to 1444.91,437(1430.0–1444.9) | 7426 seeds | Cost kind unknown |
34[30–37] MiniMaxDossier | No effort set | No harness | MiniMax M3, Arena rating: 1,437 (scale 1350.0 to 1514.8), 95% confidence interval 1428.8 to 1445.01,437(1428.8–1445.0) | 6617 seeds | Cost kind unknown |
35[32–37] GoogleDossier | No effort set | No harness | Gemma 4 31B, Arena rating: 1,434 (scale 1350.0 to 1514.8), 95% confidence interval 1427.6 to 1441.11,434(1427.6–1441.1) | 15431 seeds | Cost kind unknown |
36[31–37] GoogleDossier | No effort set | No harness | Gemini 3 Pro Preview, Arena rating: 1,434 (scale 1350.0 to 1514.8), 95% confidence interval 1425.1 to 1442.61,434(1425.1–1442.6) | 11338 seeds | Cost kind unknown |
37[34–38] Moonshot AIDossier | thinking | No harness | Kimi K2.5, Arena rating: 1,431 (scale 1350.0 to 1514.8), 95% confidence interval 1424.0 to 1437.41,431(1424.0–1437.4) | 22601 seeds | Cost kind unknown |
38[37–41] GoogleDossier | No effort set | No harness | Gemini 2.5 Pro, Arena rating: 1,423 (scale 1350.0 to 1514.8), 95% confidence interval 1416.7 to 1428.91,423(1416.7–1428.9) | 26635 seeds | Cost kind unknown |
39[38–42] AnthropicDossier | No effort set | No harness | Haiku 4.5, Arena rating: 1,420 (scale 1350.0 to 1514.8), 95% confidence interval 1415.1 to 1425.71,420(1415.1–1425.7) | 39423 seeds | Cost kind unknown |
40[38–42] Z.aiDossier | No effort set | No harness | GLM-5V-Turbo, Arena rating: 1,419 (scale 1350.0 to 1514.8), 95% confidence interval 1411.4 to 1426.21,419(1411.4–1426.2) | 9136 seeds | Cost kind unknown |
41[38–42] xAIDossier | thinking | No harness | Grok 4.20 0309, Arena rating: 1,417 (scale 1350.0 to 1514.8), 95% confidence interval 1410.2 to 1424.01,417(1410.2–1424.0) | 21926 seeds | Cost kind unknown |
42[38–44] GoogleDossier | No effort set | No harness | Gemini 3 Flash Preview, Arena rating: 1,414 (scale 1350.0 to 1514.8), 95% confidence interval 1405.0 to 1423.61,414(1405.0–1423.6) | 7538 seeds | Cost kind unknown |
43[42–46] OpenAIDossier | high | No harness | GPT-5.2, Arena rating: 1,404 (scale 1350.0 to 1514.8), 95% confidence interval 1394.7 to 1413.11,404(1394.7–1413.1) | 7552 seeds | Cost kind unknown |
44[43–46] OpenAIDossier | No effort set | No harness | GPT-5.5 Instant, Arena rating: 1,403 (scale 1350.0 to 1514.8), 95% confidence interval 1395.4 to 1411.11,403(1395.4–1411.1) | 8834 seeds | Cost kind unknown |
45[42–46] OpenAIDossier | No effort set | No harness | GPT-5.1, Arena rating: 1,403 (scale 1350.0 to 1514.8), 95% confidence interval 1393.9 to 1412.11,403(1393.9–1412.1) | 8736 seeds | Cost kind unknown |
46[43–46] OpenAIDossier | No effort set | No harness | GPT-5.2, Arena rating: 1,402 (scale 1350.0 to 1514.8), 95% confidence interval 1396.3 to 1407.91,402(1396.3–1407.9) | 29726 seeds | Cost kind unknown |
- 1AnthropicDossier
max
Arena rating
Fable 5.1, Arena rating: 1,515 (scale 1350.0 to 1514.8), 95% confidence interval 1499.7 to 1530.01,515(1499.7–1530.0) Arena rating
Opus 5, Arena rating: 1,509 (scale 1350.0 to 1514.8), 95% confidence interval 1503.0 to 1514.31,509(1503.0–1514.3)Arena rating
Opus 4.6, Arena rating: 1,508 (scale 1350.0 to 1514.8), 95% confidence interval 1501.2 to 1514.01,508(1501.2–1514.0)Arena rating
Opus 4.6, Arena rating: 1,506 (scale 1350.0 to 1514.8), 95% confidence interval 1501.0 to 1511.41,506(1501.0–1511.4)- 5AnthropicDossier
high
Arena rating
Fable 5, Arena rating: 1,497 (scale 1350.0 to 1514.8), 95% confidence interval 1490.0 to 1503.31,497(1490.0–1503.3) - 6AnthropicDossier
high
Arena rating
Opus 4.7, Arena rating: 1,496 (scale 1350.0 to 1514.8), 95% confidence interval 1489.3 to 1502.11,496(1489.3–1502.1) - 7AnthropicDossier
Arena rating
Opus 4.7, Arena rating: 1,495 (scale 1350.0 to 1514.8), 95% confidence interval 1488.4 to 1501.21,495(1488.4–1501.2) - 8OpenAIDossier
high
Arena rating
GPT-5.5, Arena rating: 1,484 (scale 1350.0 to 1514.8), 95% confidence interval 1478.9 to 1489.61,484(1478.9–1489.6) - 9GoogleDossier
high
Arena rating
Gemini 3.8 Flash, Arena rating: 1,482 (scale 1350.0 to 1514.8), 95% confidence interval 1474.0 to 1489.91,482(1474.0–1489.9) - 10AnthropicDossier
Arena rating
Sonnet 4.6, Arena rating: 1,482 (scale 1350.0 to 1514.8), 95% confidence interval 1476.0 to 1487.71,482(1476.0–1487.7) - 11OpenAIDossier
Arena rating
GPT-5.5, Arena rating: 1,481 (scale 1350.0 to 1514.8), 95% confidence interval 1475.6 to 1486.21,481(1475.6–1486.2) - 12OpenAIDossier
xhigh
Arena rating
GPT-5.6 Sol, Arena rating: 1,480 (scale 1350.0 to 1514.8), 95% confidence interval 1473.2 to 1486.81,480(1473.2–1486.8) - 13GoogleDossier
high
Arena rating
Gemini 3.7 Flash, Arena rating: 1,476 (scale 1350.0 to 1514.8), 95% confidence interval 1468.9 to 1484.11,476(1468.9–1484.1) - 14AnthropicDossier
high
Arena rating
Opus 4.8, Arena rating: 1,475 (scale 1350.0 to 1514.8), 95% confidence interval 1467.7 to 1481.41,475(1467.7–1481.4) - 15OpenAIDossier
Arena rating
GPT-5.4, Arena rating: 1,473 (scale 1350.0 to 1514.8), 95% confidence interval 1467.0 to 1478.91,473(1467.0–1478.9) - 16MetaDossier
max
Arena rating
Muse Spark 1.3, Arena rating: 1,473 (scale 1350.0 to 1514.8), 95% confidence interval 1465.1 to 1480.71,473(1465.1–1480.7) - 17OpenAIDossier
xhigh
Arena rating
GPT-5.6 Terra, Arena rating: 1,468 (scale 1350.0 to 1514.8), 95% confidence interval 1461.2 to 1474.11,468(1461.2–1474.1) - 18OpenAIDossier
max
Arena rating
GPT-6 Astra, Arena rating: 1,467 (scale 1350.0 to 1514.8), 95% confidence interval 1458.7 to 1475.81,467(1458.7–1475.8) - 19AnthropicDossier
Arena rating
Opus 4.5, Arena rating: 1,466 (scale 1350.0 to 1514.8), 95% confidence interval 1455.5 to 1475.91,466(1455.5–1475.9) - 20AnthropicDossier
Arena rating
Opus 4.8, Arena rating: 1,465 (scale 1350.0 to 1514.8), 95% confidence interval 1458.7 to 1471.41,465(1458.7–1471.4) - 21MetaDossier
Arena rating
Muse Spark 1.1, Arena rating: 1,463 (scale 1350.0 to 1514.8), 95% confidence interval 1456.4 to 1470.61,463(1456.4–1470.6) - 22GoogleDossier
high
Arena rating
Gemini 3.5 Flash, Arena rating: 1,462 (scale 1350.0 to 1514.8), 95% confidence interval 1452.6 to 1470.91,462(1452.6–1470.9) - 23GoogleDossier
medium
Arena rating
Gemini 3.5 Flash, Arena rating: 1,461 (scale 1350.0 to 1514.8), 95% confidence interval 1452.9 to 1468.41,461(1452.9–1468.4) - 24AnthropicDossier
high
Arena rating
Sonnet 5, Arena rating: 1,460 (scale 1350.0 to 1514.8), 95% confidence interval 1453.0 to 1466.11,460(1453.0–1466.1) - 25GoogleDossier
high
Arena rating
Gemini 3.6 Flash, Arena rating: 1,456 (scale 1350.0 to 1514.8), 95% confidence interval 1445.4 to 1466.81,456(1445.4–1466.8) - 26OpenAIDossier
xhigh
Arena rating
GPT-5.6 Luna, Arena rating: 1,454 (scale 1350.0 to 1514.8), 95% confidence interval 1446.8 to 1460.81,454(1446.8–1460.8) - 27xAIDossier
Arena rating
Grok 4.5, Arena rating: 1,451 (scale 1350.0 to 1514.8), 95% confidence interval 1443.9 to 1459.11,451(1443.9–1459.1) - 28Moonshot AIDossier
Arena rating
Kimi K2.6, Arena rating: 1,450 (scale 1350.0 to 1514.8), 95% confidence interval 1442.9 to 1458.01,450(1442.9–1458.0) - 29AnthropicDossier
Arena rating
Sonnet 4.5, Arena rating: 1,449 (scale 1350.0 to 1514.8), 95% confidence interval 1443.3 to 1455.41,449(1443.3–1455.4) - 30MetaDossier
Arena rating
Muse Spark, Arena rating: 1,444 (scale 1350.0 to 1514.8), 95% confidence interval 1425.4 to 1461.71,444(1425.4–1461.7) - 31GoogleDossier
Arena rating
Gemini 3.1 Pro Preview, Arena rating: 1,443 (scale 1350.0 to 1514.8), 95% confidence interval 1438.1 to 1448.41,443(1438.1–1448.4) - 32xAIDossier
high
Arena rating
Grok 4.6, Arena rating: 1,438 (scale 1350.0 to 1514.8), 95% confidence interval 1430.3 to 1445.81,438(1430.3–1445.8) - 33AlibabaDossier
Arena rating
Qwen3.7 Plus, Arena rating: 1,437 (scale 1350.0 to 1514.8), 95% confidence interval 1430.0 to 1444.91,437(1430.0–1444.9) - 34MiniMaxDossier
Arena rating
MiniMax M3, Arena rating: 1,437 (scale 1350.0 to 1514.8), 95% confidence interval 1428.8 to 1445.01,437(1428.8–1445.0) - 35GoogleDossier
Arena rating
Gemma 4 31B, Arena rating: 1,434 (scale 1350.0 to 1514.8), 95% confidence interval 1427.6 to 1441.11,434(1427.6–1441.1) - 36GoogleDossier
Arena rating
Gemini 3 Pro Preview, Arena rating: 1,434 (scale 1350.0 to 1514.8), 95% confidence interval 1425.1 to 1442.61,434(1425.1–1442.6) - 37Moonshot AIDossier
thinking
Arena rating
Kimi K2.5, Arena rating: 1,431 (scale 1350.0 to 1514.8), 95% confidence interval 1424.0 to 1437.41,431(1424.0–1437.4) - 38GoogleDossier
Arena rating
Gemini 2.5 Pro, Arena rating: 1,423 (scale 1350.0 to 1514.8), 95% confidence interval 1416.7 to 1428.91,423(1416.7–1428.9) - 39AnthropicDossier
Arena rating
Haiku 4.5, Arena rating: 1,420 (scale 1350.0 to 1514.8), 95% confidence interval 1415.1 to 1425.71,420(1415.1–1425.7) - 40Z.aiDossier
Arena rating
GLM-5V-Turbo, Arena rating: 1,419 (scale 1350.0 to 1514.8), 95% confidence interval 1411.4 to 1426.21,419(1411.4–1426.2) - 41xAIDossier
thinking
Arena rating
Grok 4.20 0309, Arena rating: 1,417 (scale 1350.0 to 1514.8), 95% confidence interval 1410.2 to 1424.01,417(1410.2–1424.0) - 42GoogleDossier
Arena rating
Gemini 3 Flash Preview, Arena rating: 1,414 (scale 1350.0 to 1514.8), 95% confidence interval 1405.0 to 1423.61,414(1405.0–1423.6) - 43OpenAIDossier
high
Arena rating
GPT-5.2, Arena rating: 1,404 (scale 1350.0 to 1514.8), 95% confidence interval 1394.7 to 1413.11,404(1394.7–1413.1) - 44OpenAIDossier
Arena rating
GPT-5.5 Instant, Arena rating: 1,403 (scale 1350.0 to 1514.8), 95% confidence interval 1395.4 to 1411.11,403(1395.4–1411.1) - 45OpenAIDossier
Arena rating
GPT-5.1, Arena rating: 1,403 (scale 1350.0 to 1514.8), 95% confidence interval 1393.9 to 1412.11,403(1393.9–1412.1) - 46OpenAIDossier
Arena rating
GPT-5.2, Arena rating: 1,402 (scale 1350.0 to 1514.8), 95% confidence interval 1396.3 to 1407.91,402(1396.3–1407.9)