Skip to content

Benchmark

Arena Agent Praise Complaint

Agentic benchmarks measure how well models carry out tasks that take several steps. Higher scores are better.

Models measured
62
Scores recorded
68
Sources
1
IQ weight
2.69

Score distribution

0.034
Mean
-0.015
Median
0.188
Std dev
-0.161
10th percentile
0.293
90th percentile
-0.273
Min
-0.2730.687
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.69 for chat models, measured over 62 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Claude Fable 5.2Max· LMArena data0.687
2Gemini 4 NeonHigh· LMArena data0.445
3Gemini 4 CarbonHigh· LMArena data0.394
4Claude Opus 5.5Max· LMArena data0.386
5GPT-6.2 AstraMax· LMArena data0.37
6GPT-6 AstraMax· LMArena data0.323
7Claude Sonnet 5.5Max· LMArena data0.307
8Gemini 4 HydrogenHigh· LMArena data0.293
9Claude Opus 5.5High· LMArena data0.289
10Claude Fable 5.1Max· LMArena data0.281
11GPT-6.1 SolMax· LMArena data0.245
12Gemini 4 ArgonHigh· LMArena data0.238
13GPT-6.2 SolMax· LMArena data0.233
14GPT-6.1 TerraMax· LMArena data0.179
15GPT-6.1 SolMedium· LMArena data0.157
16Claude Sonnet 5.5High· LMArena data0.146
17Muse Wonder FullMax· LMArena data0.142
18Qwen4 MaxxHigh· LMArena data0.13
19Claude Fable 5High· LMArena data0.115
20Claude Opus 5Max· LMArena data0.103
21GPT-6 SolMax· LMArena data0.0975
22Hy 4· LMArena data0.094
23Kimi K3.1 ProMax· LMArena data0.0672
24Gemini 3.8 FlashHigh· LMArena data0.0403
25GPT-5.6 SolxHigh· LMArena data0.0382
26GPT-6.1 TerraMedium· LMArena data0.0372
27GLM 5.3Max· LMArena data0.0262
28MiMo V2.6 Pro· LMArena data0.0251
29Grok 4.7xHigh· LMArena data0.0214
30GLM 5.2Max· LMArena data0.00728
31Kimi K3Max· LMArena data0.00419
32Muse Spark 1.3Max· LMArena data0.000562
33Inkling 2 Small· LMArena data-0.0114
34DeepSeek V4.1 Flash 1017Max· LMArena data-0.0153
35Hy4 preview· LMArena data-0.0162
36Gemini 3.1 Pro Preview· LMArena data-0.0209
37Qwen3.8 Max· LMArena data-0.0227
38Muse EmberMax· LMArena data-0.0269
39Muse WonderMax· LMArena data-0.0296
40GPT-5.6 TerraxHigh· LMArena data-0.033
41Claude Sonnet 5High· LMArena data-0.0331
42Grok 4.6xHigh· LMArena data-0.0349
43GPT-6 LunaMax· LMArena data-0.0368
44MiMo V2.6 Flash· LMArena data-0.0369
45Gemini 3.7 FlashHigh· LMArena data-0.0457
46Claude Haiku 5.5High· LMArena data-0.0551
47Qwen 3.8 Flash Next· LMArena data-0.0593
48GLM 5.3 Flash· LMArena data-0.0617
49Qwen 3.8 27B· LMArena data-0.0701
50Step 5 Preview· LMArena data-0.0791

Showing 1–50 of 68 results.

Data credits: LMArena, CC BY 4.0 (licence)