Dataset Viewer
Auto-converted to Parquet Duplicate
model_name
stringlengths
3
46
organization
stringclasses
19 values
license
stringclasses
9 values
index_score
float64
780
1.49k
index_score_ci_lower
float64
743
1.43k
index_score_ci_upper
float64
813
1.54k
raw_score
null
raw_score_ci_lower
null
raw_score_ci_upper
null
session_count
int64
958
176k
rank
int64
1
68
category
stringclasses
4 values
leaderboard_publish_date
stringdate
2026-11-18 00:00:00
2026-11-18 00:00:00
Claude Fable 5.2 (Max)
anthropic
Proprietary
1,447.853989
1,409.786102
1,485.921876
null
null
null
8,937
1
overall
2026-11-18
Claude Opus 5.5 (Max)
anthropic
Proprietary
1,335.506624
1,309.4951
1,361.518148
null
null
null
9,529
2
overall
2026-11-18
Gemini 4 Neon (High)
google
Proprietary
1,332.695076
1,300.344768
1,365.045383
null
null
null
4,238
3
overall
2026-11-18
Gemini 4 Carbon (High)
google
Proprietary
1,303.427665
1,294.605604
1,312.249726
null
null
null
45,045
4
overall
2026-11-18
Claude Sonnet 5.5 (Max)
anthropic
Proprietary
1,296.051022
1,258.770068
1,333.331977
null
null
null
6,029
5
overall
2026-11-18
GPT-6.2 Astra (Max)
openai
Proprietary
1,290.996428
1,255.752538
1,326.240318
null
null
null
3,208
6
overall
2026-11-18
GPT-6 Astra (Max)
openai
Proprietary
1,282.484614
1,256.375952
1,308.593276
null
null
null
11,498
7
overall
2026-11-18
Claude Opus 5.5 (High)
anthropic
Proprietary
1,272.957513
1,259.346159
1,286.568867
null
null
null
18,431
8
overall
2026-11-18
Claude Fable 5.1 (Max)
anthropic
Proprietary
1,271.602917
1,244.861651
1,298.344183
null
null
null
11,511
9
overall
2026-11-18
Gemini 4 Hydrogen (High)
google
Proprietary
1,270.564811
1,247.962125
1,293.167498
null
null
null
6,701
10
overall
2026-11-18
GPT-6.2 Sol (Max)
openai
Proprietary
1,240.759965
1,209.668054
1,271.851876
null
null
null
3,764
11
overall
2026-11-18
GPT-6.1 Sol (Max)
openai
Proprietary
1,240.707691
1,226.435401
1,254.979981
null
null
null
23,948
12
overall
2026-11-18
GPT-6.1 Terra (Max)
openai
Proprietary
1,230.125466
1,214.97979
1,245.271141
null
null
null
21,947
13
overall
2026-11-18
Gemini 4 Argon (High)
google
Proprietary
1,229.04795
1,217.230034
1,240.865866
null
null
null
38,821
14
overall
2026-11-18
Muse Wonder Full (Max)
meta
Proprietary
1,200.843436
1,189.092864
1,212.594008
null
null
null
24,794
15
overall
2026-11-18
GPT-6.1 Sol (Medium)
openai
Proprietary
1,200.812739
1,187.894888
1,213.730591
null
null
null
18,856
16
overall
2026-11-18
Claude Sonnet 5.5 (High)
anthropic
Proprietary
1,195.91957
1,171.083146
1,220.755994
null
null
null
8,640
17
overall
2026-11-18
Qwen4-Max (xHigh)
alibaba
Proprietary
1,188.813207
1,171.557073
1,206.06934
null
null
null
10,661
18
overall
2026-11-18
GPT-6 Sol (Max)
openai
Proprietary
1,178.452454
1,158.723203
1,198.181705
null
null
null
10,330
19
overall
2026-11-18
Claude Opus 5 (Max)
anthropic
Proprietary
1,176.363228
1,159.614743
1,193.111714
null
null
null
18,106
20
overall
2026-11-18
Hy4
tencent
Proprietary
1,175.96477
1,160.549205
1,191.380336
null
null
null
20,208
21
overall
2026-11-18
Claude Fable 5 (High)
anthropic
Proprietary
1,172.784593
1,157.255614
1,188.313573
null
null
null
30,111
22
overall
2026-11-18
GPT-6.1 Terra (Medium)
openai
Proprietary
1,154.579658
1,147.23667
1,161.922647
null
null
null
44,143
23
overall
2026-11-18
Kimi K3.1 Pro (Max)
moonshot
Kimi K3.1 Pro License
1,145.217629
1,130.802288
1,159.632971
null
null
null
17,754
24
overall
2026-11-18
GPT-5.6 Sol (xHigh)
openai
Proprietary
1,142.690309
1,128.469617
1,156.911002
null
null
null
28,015
25
overall
2026-11-18
Inkling 2 Small
thinky
Apache 2.0
1,118.09973
1,105.284787
1,130.914674
null
null
null
17,090
26
overall
2026-11-18
Gemini 3.8 Flash (High)
google
Proprietary
1,117.836886
1,106.421412
1,129.252359
null
null
null
35,945
27
overall
2026-11-18
MiMo-V2.6-Pro
xiaomi
MIT
1,116.914391
1,105.394106
1,128.434676
null
null
null
39,750
28
overall
2026-11-18
Grok 4.7 (xHigh)
xai
Proprietary
1,114.370773
1,094.822819
1,133.918727
null
null
null
13,851
29
overall
2026-11-18
Claude Sonnet 5 (High)
anthropic
Proprietary
1,110.432669
1,095.015228
1,125.850109
null
null
null
23,223
30
overall
2026-11-18
Muse Spark 1.3 (Max)
meta
Proprietary
1,110.113005
1,098.775677
1,121.450333
null
null
null
61,391
31
overall
2026-11-18
DeepSeek-V4.1-Flash-20261017 (Max)
deepseek
MIT
1,109.001538
1,102.733333
1,115.269743
null
null
null
76,376
32
overall
2026-11-18
Muse Ember (Max)
meta
Proprietary
1,108.521575
1,096.605035
1,120.438116
null
null
null
26,702
33
overall
2026-11-18
Muse Wonder (Max)
meta
Proprietary
1,106.684229
1,090.263283
1,123.105176
null
null
null
12,191
34
overall
2026-11-18
GPT-6 Luna (Max)
openai
Proprietary
1,103.567075
1,097.474723
1,109.659426
null
null
null
175,945
35
overall
2026-11-18
Claude Haiku 5.5 (High)
anthropic
Proprietary
1,095.266424
1,082.059186
1,108.473662
null
null
null
25,927
36
overall
2026-11-18
GLM-5.3 (Max)
zai
MIT
1,094.788144
1,083.619479
1,105.956809
null
null
null
63,856
37
overall
2026-11-18
Hy4 preview
tencent
Apache 2.0
1,093.70198
1,084.247186
1,103.156774
null
null
null
66,643
38
overall
2026-11-18
Kimi K3 (Max)
moonshot
Kimi K3 license
1,093.429052
1,085.171633
1,101.686471
null
null
null
98,926
39
overall
2026-11-18
MiMo-V2.6-Flash
xiaomi
MIT
1,091.823648
1,080.094116
1,103.553181
null
null
null
27,541
40
overall
2026-11-18
GLM-5.2 (Max)
zai
MIT
1,090.066176
1,079.355317
1,100.777036
null
null
null
53,427
41
overall
2026-11-18
Qwen3.8-Max
alibaba
Proprietary
1,089.743259
1,080.741483
1,098.745035
null
null
null
58,378
42
overall
2026-11-18
Grok 4.6 (xHigh)
xai
Proprietary
1,087.66311
1,070.984913
1,104.341306
null
null
null
20,845
43
overall
2026-11-18
GPT-5.6 Terra (xHigh)
openai
Proprietary
1,061.025306
1,050.216659
1,071.833952
null
null
null
30,460
44
overall
2026-11-18
Gemini 3.7 Flash (High)
google
Proprietary
1,057.010621
1,044.945826
1,069.075417
null
null
null
44,425
45
overall
2026-11-18
GPT-6 Luna (Medium)
openai
Proprietary
1,056.094295
1,050.469954
1,061.718637
null
null
null
77,636
46
overall
2026-11-18
MiniMax-M3.1-Flash
minimax
MiniMax Community License
1,050.228073
1,031.644629
1,068.811517
null
null
null
11,224
47
overall
2026-11-18
GLM-5.3-Flash
zai
MIT
1,045.633757
1,036.078716
1,055.188799
null
null
null
70,746
48
overall
2026-11-18
Step 5 Preview
stepfun
Proprietary
1,043.316233
1,025.29365
1,061.338815
null
null
null
12,692
49
overall
2026-11-18
Qwen3.8-Flash-Next
alibaba
qwen-community-1.0
1,038.646703
1,028.631896
1,048.661509
null
null
null
72,871
50
overall
2026-11-18
Gemini 3.1 Pro Preview
google
Proprietary
1,030.517274
1,010.037971
1,050.996577
null
null
null
42,107
51
overall
2026-11-18
Qwen3.8-27B
alibaba
Apache 2.0
1,028.112696
1,016.074749
1,040.150643
null
null
null
45,064
52
overall
2026-11-18
GPT-6 Luna (Low)
openai
Proprietary
1,013.204078
1,004.324319
1,022.083837
null
null
null
39,242
53
overall
2026-11-18
Beam 1.0 501B A23B
reflection
Apache 2.0
1,009.790225
976.381045
1,043.199405
null
null
null
9,403
54
overall
2026-11-18
Mistral Large 4
mistral
Modified MIT
1,006.323024
980.212836
1,032.433213
null
null
null
4,540
55
overall
2026-11-18
Hy3
tencent
Apache 2.0
1,000.301738
981.556145
1,019.04733
null
null
null
18,433
56
overall
2026-11-18
Qwen3.7-Max
alibaba
Proprietary
993.962476
977.346372
1,010.57858
null
null
null
23,989
57
overall
2026-11-18
Qwen3.7-Plus
alibaba
Proprietary
986.929613
966.533278
1,007.325947
null
null
null
14,086
58
overall
2026-11-18
Muse Spark 1.1
meta
Proprietary
979.055673
970.381163
987.730183
null
null
null
91,110
59
overall
2026-11-18
Muse Spark 1.2 (xHigh)
meta
Proprietary
978.696686
966.868523
990.52485
null
null
null
39,733
60
overall
2026-11-18
MiniMax-M3
minimax
MiniMax Community License
959.546014
944.996228
974.0958
null
null
null
25,305
61
overall
2026-11-18
Spark-X2.6
synlan
Proprietary
951.489222
937.882207
965.096238
null
null
null
12,273
62
overall
2026-11-18
Mistral Medium 3.5
mistral
Modified MIT
928.262164
906.862331
949.661996
null
null
null
9,042
63
overall
2026-11-18
NVIDIA-Nemotron-3.5-Super-VL-120B-A12B (NVFP4)
nvidia
OpenMDW-1.1
902.131035
893.550926
910.711143
null
null
null
32,809
64
overall
2026-11-18
Nemotron 3 Ultra
nvidia
OpenMDW-1.1
883.934411
840.954037
926.914785
null
null
null
4,671
65
overall
2026-11-18
Inkling
thinky
Apache 2.0
871.621666
852.643903
890.599429
null
null
null
22,774
66
overall
2026-11-18
Inkling-Small
thinky
Apache 2.0
868.916846
843.745181
894.088512
null
null
null
9,228
67
overall
2026-11-18
Solar Pro 4
upstage
Proprietary
800.030532
771.123991
828.937073
null
null
null
7,714
68
overall
2026-11-18
Claude Fable 5.2 (Max)
anthropic
Proprietary
1,368.118475
1,294.126964
1,442.109987
null
null
null
2,600
1
chat
2026-11-18
Claude Opus 5.5 (Max)
anthropic
Proprietary
1,340.720364
1,285.067321
1,396.373407
null
null
null
2,731
2
chat
2026-11-18
Gemini 4 Neon (High)
google
Proprietary
1,322.467823
1,252.294725
1,392.64092
null
null
null
1,132
3
chat
2026-11-18
Gemini 4 Carbon (High)
google
Proprietary
1,298.684471
1,279.243215
1,318.125727
null
null
null
12,048
4
chat
2026-11-18
Claude Sonnet 5.5 (Max)
anthropic
Proprietary
1,298.324913
1,213.831219
1,382.818607
null
null
null
1,741
5
chat
2026-11-18
Gemini 4 Hydrogen (High)
google
Proprietary
1,291.945511
1,241.267596
1,342.623426
null
null
null
1,833
6
chat
2026-11-18
Claude Fable 5.1 (Max)
anthropic
Proprietary
1,265.493704
1,218.696272
1,312.291136
null
null
null
4,214
7
chat
2026-11-18
GPT-6 Astra (Max)
openai
Proprietary
1,265.469716
1,213.170595
1,317.768837
null
null
null
3,739
8
chat
2026-11-18
Gemini 4 Argon (High)
google
Proprietary
1,244.128665
1,218.317565
1,269.939765
null
null
null
10,408
9
chat
2026-11-18
Claude Opus 5.5 (High)
anthropic
Proprietary
1,240.594376
1,215.32597
1,265.862782
null
null
null
6,288
10
chat
2026-11-18
GPT-6.2 Astra (Max)
openai
Proprietary
1,210.803068
1,145.57966
1,276.026477
null
null
null
958
11
chat
2026-11-18
GPT-6.1 Sol (Max)
openai
Proprietary
1,206.333449
1,178.246142
1,234.420756
null
null
null
7,088
12
chat
2026-11-18
GPT-6.1 Terra (Max)
openai
Proprietary
1,193.216861
1,161.784764
1,224.648958
null
null
null
6,505
13
chat
2026-11-18
Muse Wonder Full (Max)
meta
Proprietary
1,177.970711
1,153.497789
1,202.443633
null
null
null
7,219
14
chat
2026-11-18
Claude Opus 5 (Max)
anthropic
Proprietary
1,170.226467
1,134.876
1,205.576934
null
null
null
6,156
15
chat
2026-11-18
GPT-6.2 Sol (Max)
openai
Proprietary
1,167.023465
1,109.158882
1,224.888047
null
null
null
1,176
16
chat
2026-11-18
Qwen4-Max (xHigh)
alibaba
Proprietary
1,165.749531
1,129.369168
1,202.129895
null
null
null
3,149
17
chat
2026-11-18
GPT-6.1 Sol (Medium)
openai
Proprietary
1,165.484699
1,137.964764
1,193.004635
null
null
null
5,271
18
chat
2026-11-18
Claude Fable 5 (High)
anthropic
Proprietary
1,160.04325
1,132.740296
1,187.346203
null
null
null
11,287
19
chat
2026-11-18
Gemini 3.8 Flash (High)
google
Proprietary
1,154.455651
1,132.117288
1,176.794014
null
null
null
11,697
20
chat
2026-11-18
GPT-5.6 Sol (xHigh)
openai
Proprietary
1,142.056494
1,113.14217
1,170.970818
null
null
null
10,118
21
chat
2026-11-18
Claude Sonnet 5.5 (High)
anthropic
Proprietary
1,136.172761
1,090.122156
1,182.223365
null
null
null
2,736
22
chat
2026-11-18
Hy4
tencent
Proprietary
1,132.227753
1,098.168702
1,166.286803
null
null
null
5,706
23
chat
2026-11-18
GPT-6 Sol (Max)
openai
Proprietary
1,123.980306
1,086.029141
1,161.931471
null
null
null
3,273
24
chat
2026-11-18
GPT-6.1 Terra (Medium)
openai
Proprietary
1,123.145472
1,107.666077
1,138.624867
null
null
null
12,626
25
chat
2026-11-18
Kimi K3.1 Pro (Max)
moonshot
Kimi K3.1 Pro License
1,104.208313
1,074.339245
1,134.077382
null
null
null
5,197
26
chat
2026-11-18
Muse Spark 1.3 (Max)
meta
Proprietary
1,100.413062
1,077.375387
1,123.450737
null
null
null
20,751
27
chat
2026-11-18
Grok 4.7 (xHigh)
xai
Proprietary
1,094.945657
1,053.556273
1,136.335042
null
null
null
4,263
28
chat
2026-11-18
Gemini 3.7 Flash (High)
google
Proprietary
1,090.139353
1,067.515289
1,112.763417
null
null
null
15,808
29
chat
2026-11-18
MiMo-V2.6-Pro
xiaomi
MIT
1,084.825527
1,060.40191
1,109.249145
null
null
null
11,326
30
chat
2026-11-18
Grok 4.6 (xHigh)
xai
Proprietary
1,081.137028
1,050.490527
1,111.78353
null
null
null
7,444
31
chat
2026-11-18
DeepSeek-V4.1-Flash-20261017 (Max)
deepseek
MIT
1,080.805398
1,066.996628
1,094.614167
null
null
null
21,642
32
chat
2026-11-18
End of preview. Expand in Data Studio

Arena Leaderboard Dataset in Lance

This release preserves 2,806,760 published leaderboard rows across all 26 configurations and both source splits. It includes the publisher’s historical snapshots (full) and the latest snapshot for each configuration as of the pinned source revision (latest). Original Arena leaderboard dataset

Data and splits

Path Contents
data/<configuration>/full.lance All historical leaderboard rows from the source full split
data/<configuration>/latest.lance The exact source latest split at the pinned revision
search/model_lookup.lance 68,189 distinct model, organization, configuration and category combinations
viewer/<configuration>/<split>.parquet The original Parquet files for Hugging Face Data Studio

The Lance tables preserve the source schemas, values, nulls, and row order. The 52 source tables and browser previews were verified against the pinned Parquet files. Table counts, source hashes, and encoder revision

Search model identities

The lookup table embeds the model name, organization, arena configuration, and leaderboard category. It stores one entry per distinct combination with a model name from the historical full splits. Each entry points to its most recently published row. For entries with the same date, it points to the first row in source order. source_row is the zero-based offset within data/<configuration>/full.lance. Ratings, ranks, costs, and other evaluation measurements are available in the original data tables.

The release generates normalized 256-dimensional float32 vectors with sentence-transformers/static-retrieval-mrl-en-v1 and a cosine IVF_FLAT index. The example uses the matching encoder revision and prints cosine similarity, model identifier, organization, publication date, and source row for the five closest model names in the overall text leaderboard. Similarity measures the match to the search query. Pinned encoder

pip install pylance==14.0.3 sentence-transformers==6.3.0
import lance
from sentence_transformers import SentenceTransformer

encoder = SentenceTransformer(
    "sentence-transformers/static-retrieval-mrl-en-v1",
    revision="f60985c706f192d45d218078e49e5a8b6f15283a",
    device="cpu",
    truncate_dim=256,
)
query = "GPT-4o"
vector = encoder.encode(query, normalize_embeddings=True)
lookup = lance.dataset("hf://datasets/lance-format/leaderboard-dataset-lance/search/model_lookup.lance")
hits = lookup.to_table(
    columns=["model_name", "organization", "leaderboard_publish_date", "source_row", "_distance"],
    filter="configuration = 'text' AND category = 'overall'",
    prefilter=True,
    nearest={"column": "embedding", "q": vector, "k": 5, "metric": "cosine"},
).to_pylist()
for hit in hits:
    similarity = 1.0 - hit["_distance"]
    print(f"{similarity:.4f}\t{hit['model_name']}\t{hit['organization']}\t"
          f"{hit['leaderboard_publish_date']}\ttext/full/{hit['source_row']}")

Use the returned row offsets with lance.dataset("hf://datasets/lance-format/leaderboard-dataset-lance/data/text/full.lance").take(...) to fetch the original evaluation records. Other configurations and categories are searchable through the same lookup table.

Source and license

Credit Arena (lmarena-ai) for compiling and publishing the leaderboard snapshots. The source data and this adaptation are distributed under the Creative Commons Attribution 4.0 International license. The release adds lossless Lance tables and the derived model lookup with embeddings. The publisher’s card documents the evaluation methodology, including changes to rating scales and Agent Arena metrics. Archived source card · CC BY 4.0 license · Arena leaderboard changelog

Downloads last month
-