// ranked · September 2026

Best Vision LLM

Models that accept image input, ranked by LiveBench overall score. LiveBench doesn’t score vision separately, so this ranks general capability among models that can see.

OpenRouter + LiveBenchAll rankingsAll models

Claude Fable 5.1 (83.4) and Claude Opus 5.5 (83.2) are effectively tied on LiveBench overall — less than a point apart, which effort settings alone can move. Claude Fable 5 follows at 83.0.

Claude Fable 5.1 vs Claude Opus 5.5, head to head

Top 10

#ModelOverall scoreBlended / 1MContextCapabilities
1Claude Fable 5.1

anthropic/claude-fable-5.1

83.4$20.001M
reasoningtoolsvision
2Claude Opus 5.5

anthropic/claude-opus-5.5

83.2$8.001M
reasoningtoolsvision
3Claude Fable 5

anthropic/claude-fable-5

83.0$20.001M
reasoningtoolsvision
4GPT-6 Astra

openai/gpt-6-astra

82.2$20.001.1M
reasoningtoolsvision
5Muse Spark 1.3

meta/muse-spark-1.3

81.6$2.001.0M
reasoningtoolsvision
6DeepSeek V4.1 Flash

deepseek/deepseek-v4.1-flash

81.1$0.2001.0M
reasoningtoolsvision
7GPT-5.6 Sol

openai/gpt-5.6-sol

81.1$4.001.1M
reasoningtoolsvision
8GPT-5.5

openai/gpt-5.5

80.2$11.251.1M
reasoningtoolsvision
9Claude Opus 5

anthropic/claude-opus-5

80.1$10.001M
reasoningtoolsvision
10GPT-6 Sol

openai/gpt-6-sol

79.2$4.001.1M
reasoningtoolsvision

37 more models qualify — see the full leaderboard.

FAQ

What is the best vision LLM in September 2026?

Claude Fable 5.1 (83.4) and Claude Opus 5.5 (83.2) are effectively tied on LiveBench overall — less than a point apart, which effort settings alone can move. Claude Fable 5 follows at 83.0.

How is this ranking produced?

Models that accept image input, ranked by LiveBench overall score. LiveBench doesn’t score vision separately, so this ranks general capability among models that can see.

Other rankings

How this ranking is produced

  • One metric, stated above — nothing here is weighted or scored by us. Scores come from LiveBench release 2026-06-25; models without a published run don't appear in score-based rankings.
  • Price, context and capabilities — live from OpenRouter, refreshed every 15 minutes.
  • Ties — scores less than a point apart are called a tie; effort settings alone move a LiveBench score by more than that.

A public benchmark is someone else's workload. Before committing, see LLM & agent evaluation.