Dataset Viewer
Auto-converted to Parquet Duplicate
model_name
stringlengths
3
35
organization
stringlengths
3
9
license
stringclasses
7 values
score
float64
-0.13
0.15
score_ci_lower
float64
-0.15
0.13
score_ci_upper
float64
-0.12
0.16
observation_count
float64
199k
8.44M
session_count
float64
5.77k
99.8k
rank
int64
1
43
category
stringclasses
1 value
leaderboard_publish_date
stringdate
2026-09-08 00:00:00
2026-09-08 00:00:00
Claude Fable 5.1 (Max)
anthropic
Proprietary
0.145069
0.126065
0.164073
1,093,061
11,788
1
overall
2026-09-08
GPT 6 Astra (Max)
openai
Proprietary
0.125451
0.103698
0.147204
636,411
8,929
2
overall
2026-09-08
Claude Opus 5 (High)
anthropic
Proprietary
0.114169
0.096895
0.131442
2,794,373
23,232
3
overall
2026-09-08
Claude Opus 5 (Max)
anthropic
Proprietary
0.107863
0.088631
0.127096
2,321,802
18,707
4
overall
2026-09-08
Claude Fable 5 (High)
anthropic
Proprietary
0.091527
0.07717
0.105884
1,919,570
36,723
5
overall
2026-09-08
GPT 5.6 Sol (xHigh)
openai
Proprietary
0.077499
0.063003
0.091995
3,300,495
30,450
6
overall
2026-09-08
Claude Opus 4.8 (High)
anthropic
Proprietary
0.07686
0.062474
0.091247
1,983,924
38,862
7
overall
2026-09-08
Kimi K3 (Max)
moonshot
Kimi K3 license
0.065917
0.058945
0.072889
8,436,372
99,807
8
overall
2026-09-08
Claude Sonnet 5 (High)
anthropic
Proprietary
0.063102
0.044268
0.081936
2,970,415
29,365
9
overall
2026-09-08
Hy4 preview
tencent
Apache 2.0
0.053722
0.039692
0.067752
1,502,776
17,107
10
overall
2026-09-08
GPT 5.5 (xHigh)
openai
Proprietary
0.053341
0.042525
0.064157
2,399,087
51,887
11
overall
2026-09-08
GLM 5.2 (Max)
zai
MIT
0.046397
0.038884
0.053911
4,294,268
71,824
12
overall
2026-09-08
DeepSeek V4 Pro (High) (0813)
deepseek
MIT
0.044302
0.03512
0.053484
3,157,720
37,322
13
overall
2026-09-08
Gemini 3.8 Flash (High)
google
Proprietary
0.040137
0.020525
0.059748
567,897
10,105
14
overall
2026-09-08
Grok 4.5
xai
Proprietary
0.039333
0.027951
0.050716
1,788,189
35,991
15
overall
2026-09-08
Qwen3.8 Max
alibaba
Proprietary
0.039017
0.02863
0.049404
1,741,167
22,316
16
overall
2026-09-08
Grok 4.6 (xHigh)
xai
Proprietary
0.032447
0.020731
0.044164
2,096,177
19,855
17
overall
2026-09-08
GPT 5.5
openai
Proprietary
0.029519
0.020353
0.038684
2,035,367
79,963
18
overall
2026-09-08
GLM 5.3 (Max)
zai
MIT
0.02564
0.018849
0.032431
4,985,226
60,879
19
overall
2026-09-08
Deepseek V4 Flash (High) (20260731)
deepseek
MIT
0.022176
0.014237
0.030115
6,083,220
61,199
20
overall
2026-09-08
GLM 5.3 Flash
zai
MIT
0.019962
0.010799
0.029126
2,430,955
23,765
21
overall
2026-09-08
GPT 5.4 (High)
openai
Proprietary
0.015699
0.006579
0.02482
3,597,323
79,102
22
overall
2026-09-08
GPT 5.6 Terra (xHigh)
openai
Proprietary
0.014727
0.002868
0.026585
1,105,527
18,667
23
overall
2026-09-08
GPT 5.6 Luna (xHigh)
openai
Proprietary
0.011139
0.000359
0.02192
1,729,938
21,120
24
overall
2026-09-08
Qwen3.8 Flash Next
alibaba
qwen-community-1.0
0.008071
-0.002333
0.018474
3,848,644
26,730
25
overall
2026-09-08
Qwen 3.8 27B
alibaba
Apache 2.0
0.001407
-0.007826
0.010639
2,972,540
26,260
26
overall
2026-09-08
DeepSeek V4 Pro
deepseek
MIT
-0.004243
-0.015648
0.007162
1,638,264
32,810
27
overall
2026-09-08
Claude Sonnet 4.6
anthropic
Proprietary
-0.010492
-0.023599
0.002615
1,444,598
39,399
28
overall
2026-09-08
Muse Spark 1.2 (xHigh)
meta
Proprietary
-0.015562
-0.027878
-0.003245
960,824
23,160
29
overall
2026-09-08
Muse Spark 1.1
meta
Proprietary
-0.026492
-0.032915
-0.020069
3,573,742
93,764
30
overall
2026-09-08
Qwen3.7 Max
alibaba
Proprietary
-0.030932
-0.040695
-0.021168
1,706,617
37,028
31
overall
2026-09-08
Hy3
tencent
Apache 2.0
-0.042166
-0.055294
-0.029038
916,707
25,750
32
overall
2026-09-08
Mimo V2.5 Pro
xiaomi
MIT
-0.050762
-0.060636
-0.040889
1,735,075
38,400
33
overall
2026-09-08
Qwen3.7 Plus
alibaba
Proprietary
-0.052044
-0.064904
-0.039185
955,069
20,413
34
overall
2026-09-08
Minimax M3
minimax
MiniMax Community License
-0.052581
-0.062096
-0.043067
2,722,010
37,451
35
overall
2026-09-08
Gemini 3.6 Flash (High)
google
Proprietary
-0.053029
-0.064373
-0.041685
1,070,856
19,113
36
overall
2026-09-08
Gemini 3.1 Pro Preview
google
Proprietary
-0.053429
-0.063016
-0.043841
2,844,974
85,159
37
overall
2026-09-08
Inkling Small
thinky
Apache 2.0
-0.068949
-0.085388
-0.05251
399,192
11,959
38
overall
2026-09-08
Inkling
thinky
Apache 2.0
-0.088732
-0.09954
-0.077924
1,583,056
41,509
39
overall
2026-09-08
Mistral Medium 3.5
mistral
Modified MIT
-0.093452
-0.108582
-0.078323
325,321
9,448
40
overall
2026-09-08
Solar Pro 4
upstage
Proprietary
-0.112926
-0.150398
-0.075455
199,038
5,770
41
overall
2026-09-08
Minimax M2.7
minimax
Modified MIT
-0.13015
-0.142404
-0.117896
858,207
37,513
42
overall
2026-09-08
Gemini 3.5 Flash Lite
google
Proprietary
-0.134337
-0.148744
-0.119929
670,203
23,216
43
overall
2026-09-08

Arena Leaderboard Dataset

Historical snapshots of the Arena leaderboard.

Usage

from datasets import load_dataset

# Load all historical text style control data
ds = load_dataset("lmarena-ai/leaderboard-dataset", "text_style_control", split="full")

# Load the current text style control leaderboard
ds = load_dataset("lmarena-ai/leaderboard-dataset", "text_style_control", split="latest")

# Filter to overall category
ds = load_dataset("lmarena-ai/leaderboard-dataset", "text_style_control", split="latest",
    filters=[("category", "==", "overall")]
)

# Track a specific model's rating over time
ds = load_dataset("lmarena-ai/leaderboard-dataset", "text_style_control", split="full",
    filters=[("category", "==", "overall"), ("model_name", "==", "gpt-4o-2024-05-13")],
    columns=["model_name", "rating", "rank", "leaderboard_publish_date"]
)

# Load raw (non-style-controlled) text ratings
ds = load_dataset("lmarena-ai/leaderboard-dataset", "text", split="full")

# Load the current Agent Arena leaderboard
ds = load_dataset("lmarena-ai/leaderboard-dataset", "agent", split="latest")

Subsets

Each arena is a separate subset, arenas with style control have an additional subset with a _style_control suffix .

  • text, text_style_control
  • vision, vision_style_control
  • search, search_style_control
  • document, document_style_control
  • webdev (Code Arena)
  • text_to_image
  • image_edit
  • text_to_video
  • image_to_video
  • video_edit
  • agent (Agent Arena aggregate), plus per-signal subsets: agent_bash_recovery_steps, agent_praise_complaint, agent_steerability, agent_task_outcome_explicit, agent_tool_hallucination

Splits

  • full: All historically published leaderboards
  • latest: Only the most recently published leaderboards

Notes

  • On January 9, 2024, the rating system was updated from Elo to Bradley-Terry.
  • On May 16, 2025, style control was made the default for text and vision arenas and the offset was adjusted to put the style control leaderboard on the same rating scale as non style control.
  • On July 23, 2025, frequency-based re-weighting was implemented.
  • Search and Webdev Arena data starts from their releases on the arena.ai (then lmarena.ai) on August 7 and November 12 2025 respectively.
  • Agent Arena data starts from its release on June 4, 2026.

For a complete record of all leaderboard methodology changes, see the Leaderboard Changelog.

Schema

Column Type Description
model_name string Model identifier
organization string Model creator/organization
license string Model license
rating float Arena Score
rating_lower float Lower confidence bound
rating_upper float Upper confidence bound
variance float Rating variance
vote_count int Number of battles for this model
rank int Rank within this leaderboard
category string Leaderboard category (e.g., overall, coding, math)
leaderboard_publish_date string Date that this score was published (YYYY-MM-DD)

Agent Arena subsets use IPS scores instead of Bradley-Terry:

Column Type Description
model_name string Model identifier
organization string Model creator/organization
license string Model license
score float IPS score (τ̂)
score_ci_lower float Lower 95% confidence bound
score_ci_upper float Upper 95% confidence bound
observation_count int Signal observations for this model
session_count int Distinct sessions (aggregate subset only)
rank int Rank within this leaderboard
category string Leaderboard category
leaderboard_publish_date string Date that this score was published (YYYY-MM-DD)
Downloads last month
28,582