Model Markets
Get Free API KeyAPI

LMArena Vision Arena

LMArena · vision-2026-08-27-011508720696 · Blind human preferences for multimodal responses in LMArena's vision arena.

Model Ranking

Source
#1Claude Fable 5Anthropic1,329.46rating—10,002claude-fable-5; 95% CI [1321.03632319, 1337.88661881]; votes 10002; rank 1unknownOriginal sourcewinner eligibleThird-party benchmark
#2Qwen3.8-MaxQwen1,312.86rating—7,244qwen3.8-max; 95% CI [1304.57707433, 1321.13715544]; votes 7244; rank 6unknownOriginal sourcewinner eligibleThird-party benchmark
#3Gemini 3.5 FlashGoogle DeepMind1,310.66rating—8,259gemini-3.5-flash-high; 95% CI [1302.35646187, 1318.96740220]; votes 8259; rank 8unknownOriginal sourcewinner eligibleThird-party benchmark
#4Gemini 3.6 FlashGoogle DeepMind1,301.97rating—3,816gemini-3.6-flash-high; 95% CI [1291.15288726, 1312.79182351]; votes 3816; rank 13unknownOriginal sourcewinner eligibleThird-party benchmark
#5GLM-5.3-FlashZ.ai1,296.37rating—1,389glm-5.3-flash; 95% CI [1279.58251138, 1313.15015420]; votes 1389; rank 15unknownOriginal sourcewinner eligibleThird-party benchmark
#6Gemini 3.1 ProGoogle DeepMind1,295.39rating—39,412gemini-3.1-pro-preview; 95% CI [1289.75236576, 1301.03738662]; votes 39412; rank 17unknownOriginal sourcewinner eligibleThird-party benchmark
#7Claude Opus 4.8Anthropic1,293.93rating—13,521claude-opus-4-8-high; 95% CI [1286.47351506, 1301.38306277]; votes 13521; rank 19unknownOriginal sourcewinner eligibleThird-party benchmark
#8Gemini 3 FlashGoogle DeepMind1,284.99rating—36,747gemini-3-flash; 95% CI [1279.54770336, 1290.43069010]; votes 36747; rank 24unknownOriginal sourcewinner eligibleThird-party benchmark
#9Qwen3.7-PlusQwen1,281.06rating—9,132qwen3.7-plus; 95% CI [1273.03498970, 1289.09217677]; votes 9132; rank 26unknownOriginal sourcewinner eligibleThird-party benchmark
#10Kimi-K2.6Moonshot AI1,280.71rating—15,378kimi-k2.6; 95% CI [1273.64165646, 1287.78511818]; votes 15378; rank 27unknownOriginal sourcewinner eligibleThird-party benchmark
#11Qwen3.8-27BQwen1,279.32rating—2,954qwen3.8-27b; 95% CI [1267.46655532, 1291.16933290]; votes 2954; rank 28unknownOriginal sourcewinner eligibleThird-party benchmark
#12Claude Sonnet 5Anthropic1,277.94rating—8,522claude-sonnet-5-high; 95% CI [1269.73992861, 1286.14906472]; votes 8522; rank 29unknownOriginal sourcewinner eligibleThird-party benchmark
#13GPT-5.6 SolOpenAI1,276.90rating—5,837gpt-5.6-sol-xhigh; 95% CI [1267.71657009, 1286.08047215]; votes 5837; rank 30unknownOriginal sourcewinner eligibleThird-party benchmark
#14GPT-5.6 TerraOpenAI1,268.99rating—5,774gpt-5.6-terra-xhigh; 95% CI [1259.82673197, 1278.15538857]; votes 5774; rank 33unknownOriginal sourcewinner eligibleThird-party benchmark
#15Gemini 3.5 Flash-LiteGoogle DeepMind1,268.89rating—3,886gemini-3.5-flash-lite; 95% CI [1258.10808094, 1279.66830878]; votes 3886; rank 34unknownOriginal sourcewinner eligibleThird-party benchmark
#16Kimi-K2.5Moonshot AI1,266.74rating—26,728kimi-k2.5-thinking; 95% CI [1260.83834046, 1272.64651923]; votes 26728; rank 36unknownOriginal sourcewinner eligibleThird-party benchmark
#17Qwen3.5-397B-A17BQwen1,264.80rating—25,997qwen3.5-397b-a17b; 95% CI [1258.85111028, 1270.75814841]; votes 25997; rank 38unknownOriginal sourcewinner eligibleThird-party benchmark
#18Gemini 2.5 ProGoogle DeepMind1,262.42rating—87,719gemini-2.5-pro; 95% CI [1257.72056767, 1267.11330698]; votes 87719; rank 41unknownOriginal sourcewinner eligibleThird-party benchmark
#19Grok 4.20 Multi-AgentxAI1,259.97rating—22,988grok-4.20-multi-agent-beta-0309; 95% CI [1253.56879963, 1266.36768970]; votes 22988; rank 42unknownOriginal sourcewinner eligibleThird-party benchmark
#20MiniMax-M3MiniMax1,255.42rating—13,390minimax-m3; 95% CI [1247.93588333, 1262.90862002]; votes 13390; rank 44unknownOriginal sourcewinner eligibleThird-party benchmark
#21GPT-5.6 LunaOpenAI1,254.30rating—5,857gpt-5.6-luna-xhigh; 95% CI [1245.05722338, 1263.53552757]; votes 5857; rank 45unknownOriginal sourcewinner eligibleThird-party benchmark
#22Gemini 2.5 FlashGoogle DeepMind1,235.76rating—58,968gemini-2.5-flash; 95% CI [1231.05290901, 1240.46508391]; votes 58968; rank 59unknownOriginal sourcewinner eligibleThird-party benchmark
#23Mistral Large 3Mistral AI1,226.84rating—4,751mistral-large-3; 95% CI [1216.90651393, 1236.78308197]; votes 4751; rank 66unknownOriginal sourcewinner eligibleThird-party benchmark
#24Mistral-Medium-3.5-128BMistral AI1,222.18rating—5,091mistral-medium-3.5; 95% CI [1212.39755732, 1231.95907443]; votes 5091; rank 67unknownOriginal sourcewinner eligibleThird-party benchmark
#25GPT-4.1OpenAI1,210.32rating—41,034gpt-4.1-2025-04-14; 95% CI [1203.63813827, 1217.00696914]; votes 41034; rank 70unknownOriginal sourcewinner eligibleThird-party benchmark
#26GPT-4.1 MiniOpenAI1,181.17rating—40,292gpt-4.1-mini-2025-04-14; 95% CI [1173.69296987, 1188.64000739]; votes 40292; rank 84unknownOriginal sourcewinner eligibleThird-party benchmark
#27Llama-4-Maverick-17B-128E-InstructMeta1,141.47rating—6,936llama-4-maverick-17b-128e-instruct; 95% CI [1132.42029588, 1150.51619722]; votes 6936; rank 102unknownOriginal sourcewinner eligibleThird-party benchmark
#28Llama-4-Scout-17B-16E-InstructMeta1,118.08rating—6,467llama-4-scout-17b-16e-instruct; 95% CI [1108.62969338, 1127.52426769]; votes 6467; rank 110unknownOriginal sourcewinner eligibleThird-party benchmark
#29GPT-4o MiniOpenAISelected model1,065.63rating—17,341gpt-4o-mini-2024-07-18; 95% CI [1057.40637002, 1073.85507682]; votes 17341; rank 118unknownOriginal sourcewinner eligibleThird-party benchmark
#30GPT-4oOpenAI1,064.54rating—3,376gpt-4o-2024-08-06; 95% CI [1052.10281463, 1076.97369812]; votes 3376; rank 119unknownOriginal sourcewinner eligibleThird-party benchmark
#31GPT-4.1 NanoOpenAI1,063.23rating—1,211gpt-4.1-nano-2025-04-14; 95% CI [1044.95116252, 1081.50991534]; votes 1211; rank 120unknownOriginal sourcewinner eligibleThird-party benchmark

31 models ranked by the best winner-eligible current-version score when available, otherwise the best publisher-artifact score; highest first. Observed Sep 2, 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 coverage31 published models from 148 source rows; 114 source identities remain quarantined.

Primary Evidence

Publisher Artifacts

LMArena Vision Arena methodologyLMArena · observed Sep 2, 2026LMArena Vision Arena vision-2026-08-27-011508720696 score tableLMArena · observed Sep 2, 2026

Questions

LMArena Vision Arena FAQs

What does LMArena Vision Arena measure?+

Blind human preferences for multimodal responses in LMArena's vision arena. Model Markets classifies it as a multimodal benchmark and preserves the publisher's vision-2026-08-27-011508720696 release as a distinct comparison cohort.

How are models ranked on LMArena Vision 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 Vision Arena?+

Claude Fable 5 leads the current verified table with 1,329.46 rating on vision-2026-08-27-011508720696. This is a benchmark-specific result, not a universal model-quality claim.

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

31 catalog models are published from 148 source rows. 114 source identities remain quarantined rather than guessed.

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

Raw scores should be compared only within the same benchmark version, metric, and protocol. Model Markets normalizes eligible scores only for aggregate rankings, and groups models by an identical benchmark set before ranking them. View aggregate rankings →

Why might a model be missing from LMArena Vision 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 Vision Arena leaderboard?+

The current Model Markets snapshot was observed Sep 2, 2026 from LMArena artifacts. The exact publisher source and each retained result artifact are linked on this page.

Send Feedback