Model Markets
Get Free API KeyAPI

LMArena Document Arena

LMArena · document-2026-09-13-a64ca3b12b6e · Blind human preferences for model-generated documents in LMArena.

Model Ranking

Source
#1Claude Opus 5Anthropic1,516.27rating—8,794claude-opus-5-high; 95% CI [1509.24512533, 1523.28765667]; votes 8794; rank 1unknownOriginal sourcewinner eligibleThird-party benchmark
#2Claude Fable 5.1Anthropic1,512.58rating—1,403claude-fable-5.1-max; 95% CI [1497.40364732, 1527.76498981]; votes 1403; rank 2unknownOriginal sourcewinner eligibleThird-party benchmark
#3Claude Opus 4.6Anthropic1,507.32rating—27,929claude-opus-4-6-high; 95% CI [1500.89502415, 1513.74254503]; votes 27929; rank 3unknownOriginal sourcewinner eligibleThird-party benchmark
#4Claude Fable 5Anthropic1,496.08rating—8,497claude-fable-5; 95% CI [1488.22135990, 1503.94311259]; votes 8497; rank 5unknownOriginal sourcewinner eligibleThird-party benchmark
#5Claude Opus 4.7Anthropic1,495.04rating—22,192claude-opus-4-7; 95% CI [1488.66496264, 1501.41313008]; votes 22192; rank 6unknownOriginal sourcewinner eligibleThird-party benchmark
#6GPT-5.5OpenAI1,485.72rating—21,279gpt-5.5; 95% CI [1479.48245578, 1491.95617143]; votes 21279; rank 8unknownOriginal sourcewinner eligibleThird-party benchmark
#7GPT-5.6 SolOpenAI1,482.71rating—4,818gpt-5.6-sol-xhigh; 95% CI [1473.91086026, 1491.51379085]; votes 4818; rank 10unknownOriginal sourcewinner eligibleThird-party benchmark
#8Claude Sonnet 4.6Anthropic1,482.24rating—57,813claude-sonnet-4-6; 95% CI [1476.36427265, 1488.10617859]; votes 57813; rank 11unknownOriginal sourcewinner eligibleThird-party benchmark
#9Claude Opus 4.8Anthropic1,474.79rating—11,551claude-opus-4-8-high; 95% CI [1467.88529316, 1481.69160653]; votes 11551; rank 12unknownOriginal sourcewinner eligibleThird-party benchmark
#10GPT-5.6 TerraOpenAI1,471.58rating—5,787gpt-5.6-terra-xhigh; 95% CI [1463.31179298, 1479.85049651]; votes 5787; rank 13unknownOriginal sourcewinner eligibleThird-party benchmark
#11GPT-5.4OpenAI1,471.00rating—33,331gpt-5.4; 95% CI [1464.96841782, 1477.04032901]; votes 33331; rank 14unknownOriginal sourcewinner eligibleThird-party benchmark
#12Muse Spark 1.3Meta1,470.66rating—1,006muse-spark-1.3-max; 95% CI [1452.36505006, 1488.94793707]; votes 1006; rank 15unknownOriginal sourcewinner eligibleThird-party benchmark
#13GPT-6 AstraOpenAI1,467.52rating—1,621gpt-6-astra-max; 95% CI [1453.25995337, 1481.78626802]; votes 1621; rank 16unknownOriginal sourcewinner eligibleThird-party benchmark
#14Claude Sonnet 5Anthropic1,466.14rating—7,411claude-sonnet-5-high; 95% CI [1458.49407458, 1473.77896309]; votes 7411; rank 17unknownOriginal sourcewinner eligibleThird-party benchmark
#15Muse Spark 1.1Meta1,464.70rating—4,885muse-spark-1.1; 95% CI [1455.86175356, 1473.53929177]; votes 4885; rank 18unknownOriginal sourcewinner eligibleThird-party benchmark
#16Gemini 3.5 FlashGoogle DeepMind1,463.27rating—6,500gemini-3.5-flash-medium; 95% CI [1455.40466041, 1471.14299590]; votes 6500; rank 20unknownOriginal sourcewinner eligibleThird-party benchmark
#17Claude Opus 4.5Anthropic1,462.20rating—7,981claude-opus-4-5-20251101; 95% CI [1451.78988580, 1472.60655272]; votes 7981; rank 21unknownOriginal sourcewinner eligibleThird-party benchmark
#18GPT-5.6 LunaOpenAI1,456.92rating—5,766gpt-5.6-luna-xhigh; 95% CI [1448.72029252, 1465.11301394]; votes 5766; rank 23unknownOriginal sourcewinner eligibleThird-party benchmark
#19Gemini 3.6 FlashGoogle DeepMindSelected model1,456.06rating—2,975gemini-3.6-flash-high; 95% CI [1445.28023002, 1466.83474253]; votes 2975; rank 24unknownOriginal sourcewinner eligibleThird-party benchmark
#20Grok 4.6xAI1,452.33rating—2,258grok-4.6-high; 95% CI [1440.31399734, 1464.34557688]; votes 2258; rank 25unknownOriginal sourcewinner eligibleThird-party benchmark
#21Grok 4.5xAI1,451.96rating—4,965grok-4.5; 95% CI [1443.26182849, 1460.65626012]; votes 4965; rank 26unknownOriginal sourcewinner eligibleThird-party benchmark
#22Kimi K2.6Moonshot AI1,450.74rating—11,291kimi-k2.6; 95% CI [1443.10467262, 1458.37474735]; votes 11291; rank 27unknownOriginal sourcewinner eligibleThird-party benchmark
#23Claude Sonnet 4.5Anthropic1,449.81rating—31,455claude-sonnet-4-5-20250929; 95% CI [1443.68972646, 1455.92640287]; votes 31455; rank 28unknownOriginal sourcewinner eligibleThird-party benchmark
#24Qwen3.7 PlusQwen1,444.14rating—4,591qwen3.7-plus; 95% CI [1435.12302267, 1453.16578005]; votes 4591; rank 30unknownOriginal sourcewinner eligibleThird-party benchmark
#25Gemini 3.1 ProGoogle DeepMind1,443.92rating—49,457gemini-3.1-pro-preview; 95% CI [1438.69417039, 1449.14324048]; votes 49457; rank 31unknownOriginal sourcewinner eligibleThird-party benchmark
#26MiniMax M3MiniMax1,434.79rating—6,314minimax-m3; 95% CI [1426.53255805, 1443.05504941]; votes 6314; rank 32unknownOriginal sourcewinner eligibleThird-party benchmark
#27Kimi K2.5Moonshot AI1,430.10rating—21,569kimi-k2.5-thinking; 95% CI [1423.37027866, 1436.81983378]; votes 21569; rank 34unknownOriginal sourcewinner eligibleThird-party benchmark
#28Gemma 4 31BGoogle DeepMind1,424.51rating—12,326gemma-4-31b; 95% CI [1416.71159012, 1432.31286462]; votes 12326; rank 35unknownOriginal sourcewinner eligibleThird-party benchmark
#29Gemini 2.5 ProGoogle DeepMind1,421.23rating—25,110gemini-2.5-pro; 95% CI [1414.98072895, 1427.48825137]; votes 25110; rank 36unknownOriginal sourcewinner eligibleThird-party benchmark
#30Claude Haiku 4.5Anthropic1,419.97rating—34,677claude-haiku-4-5-20251001; 95% CI [1414.09579972, 1425.83505080]; votes 34677; rank 37unknownOriginal sourcewinner eligibleThird-party benchmark
#31Gemini 3 FlashGoogle DeepMind1,412.99rating—7,158gemini-3-flash; 95% CI [1403.58694998, 1422.40270038]; votes 7158; rank 40unknownOriginal sourcewinner eligibleThird-party benchmark
#32GPT-5.2OpenAI1,404.59rating—7,108gpt-5.2-high; 95% CI [1395.08956216, 1414.09210801]; votes 7108; rank 41unknownOriginal sourcewinner eligibleThird-party benchmark
#33GPT-5.1OpenAI1,403.17rating—8,244gpt-5.1; 95% CI [1393.82886201, 1412.50542979]; votes 8244; rank 42unknownOriginal sourcewinner eligibleThird-party benchmark

33 models ranked by the best winner-eligible current-version score when available, otherwise the best publisher-artifact score; highest first. Observed Oct 19, 2026. The model you came from is highlighted.

Methodology and Coverage

Ranking protocolarena rating · rating. Each model appears once; verified winner-eligible runs take precedence, and missing models are not imputed.
Acquisition coverage33 published models from 44 source rows; 5 source identities remain quarantined.

Primary Evidence

Publisher Artifacts

LMArena Vision Arena methodologyLMArena · observed Oct 19, 2026LMArena Document Arena document-2026-09-13-a64ca3b12b6e score tableLMArena · observed Oct 19, 2026

Questions

LMArena Document Arena FAQs

What does LMArena Document Arena measure?+

Blind human preferences for model-generated documents in LMArena. Model Markets classifies it as a knowledge work benchmark and preserves the publisher's document-2026-09-13-a64ca3b12b6e release as a distinct comparison cohort.

How are models ranked on LMArena Document Arena?+

Models are ordered by arena rating in rating, with higher scores ranked first. Each model appears once; a verified winner-eligible current-version result takes precedence over a publisher-artifact-only result, and missing scores are not estimated.

Which model currently leads LMArena Document Arena?+

Claude Opus 5 leads the current verified table with 1,516.27 rating on document-2026-09-13-a64ca3b12b6e. This is a benchmark-specific result, not a universal model-quality claim.

How many models have a published LMArena Document Arena score?+

33 catalog models are published from 44 source rows. 5 source identities remain quarantined rather than guessed.

Can LMArena Document Arena scores be compared with other benchmarks?+

Raw scores should be compared only within the same benchmark version, metric, and protocol. The Intelligence ranking normalizes its three pinned publisher inputs to a frozen reference range, requires at least one source, and labels one- and two-source rows provisional instead of estimating missing inputs. View aggregate rankings →

Why might a model be missing from LMArena Document Arena?+

A model remains absent when the publisher has no current result, the source model identity is unresolved, the evaluation protocol is incompatible, or the evidence cannot be verified. Model Markets does not substitute a provider claim or infer a score from a related model.

How current is the LMArena Document Arena leaderboard?+

The current Model Markets snapshot was observed Oct 19, 2026 from LMArena artifacts. The exact publisher source and each retained result artifact are linked on this page.

Send Feedback