Dataset Viewer
Auto-converted to Parquet Duplicate
model_name
stringlengths
7
29
organization
stringclasses
6 values
license
stringclasses
3 values
rating
float64
1.41k
1.53k
rating_lower
float64
1.4k
1.52k
rating_upper
float64
1.41k
1.54k
variance
float64
11.5
97.9
vote_count
int64
842
23.1k
rank
int64
1
23
category
stringclasses
1 value
leaderboard_publish_date
stringdate
2026-04-27 00:00:00
2026-04-27 00:00:00
claude-opus-4-6-thinking
anthropic
Proprietary
1,525.542366
1,516.082675
1,535.002058
23.294735
8,848
1
overall
2026-04-27
claude-opus-4-6
anthropic
Proprietary
1,520.199156
1,512.133794
1,528.264518
16.93369
15,731
2
overall
2026-04-27
claude-opus-4-7
anthropic
Proprietary
1,514.522543
1,503.33185
1,525.713236
32.600013
3,310
3
overall
2026-04-27
claude-opus-4-7-thinking
anthropic
Proprietary
1,510.747795
1,499.636709
1,521.858881
32.137852
3,179
4
overall
2026-04-27
claude-sonnet-4-6
anthropic
Proprietary
1,499.523764
1,491.998347
1,507.049181
14.742289
23,066
5
overall
2026-04-27
gpt-5.5
openai
Proprietary
1,489.916504
1,473.50494
1,506.328068
70.113841
1,145
6
overall
2026-04-27
gpt-5.5-high
openai
Proprietary
1,487.109183
1,471.222553
1,502.995814
65.70031
1,205
7
overall
2026-04-27
gpt-5.4
openai
Proprietary
1,479.691772
1,471.128118
1,488.255425
19.090706
10,655
8
overall
2026-04-27
claude-opus-4-5-20251101
anthropic
Proprietary
1,469.543936
1,459.489224
1,479.598648
26.317408
8,020
9
overall
2026-04-27
kimi-k2.6
moonshot
Modified MIT
1,456.672728
1,442.04155
1,471.303906
55.726579
1,524
10
overall
2026-04-27
muse-spark
meta
Proprietary
1,456.627754
1,437.239101
1,476.016408
97.858629
842
11
overall
2026-04-27
claude-sonnet-4-5-20250929
anthropic
Proprietary
1,449.71
1,441.368795
1,458.051205
18.111791
13,688
12
overall
2026-04-27
gemini-3.1-pro-preview
google
Proprietary
1,448.603941
1,441.950133
1,455.257748
11.525088
19,099
13
overall
2026-04-27
kimi-k2.5-thinking
moonshot
Modified MIT
1,443.539716
1,434.648867
1,452.430565
20.577391
7,811
14
overall
2026-04-27
gemini-3-pro
google
Proprietary
1,441.853265
1,433.309913
1,450.396617
19.0003
10,776
15
overall
2026-04-27
gemma-4-31b
google
Apache 2.0
1,431.568559
1,419.513399
1,443.623719
37.83117
2,835
16
overall
2026-04-27
gemini-2.5-pro
google
Proprietary
1,429.386806
1,422.003427
1,436.770185
14.191037
16,533
17
overall
2026-04-27
grok-4.20-beta-0309-reasoning
xai
Proprietary
1,425.931186
1,414.877319
1,436.985052
31.8077
3,820
18
overall
2026-04-27
claude-haiku-4-5-20251001
anthropic
Proprietary
1,423.566288
1,415.761886
1,431.370691
15.855617
14,707
19
overall
2026-04-27
gemini-3-flash
google
Proprietary
1,420.788521
1,411.757065
1,429.819977
21.233391
7,205
20
overall
2026-04-27
gpt-5.2-high
openai
Proprietary
1,413.553729
1,404.365123
1,422.742334
21.978751
7,110
21
overall
2026-04-27
gpt-5.1
openai
Proprietary
1,409.912785
1,400.863968
1,418.961601
21.315099
8,275
22
overall
2026-04-27
gpt-5.2
openai
Proprietary
1,406.453755
1,399.480609
1,413.4269
12.657888
18,730
23
overall
2026-04-27

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")

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

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.

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)
Downloads last month
8,636