Claude Opus 5.5 leads modelgrep’s vision rankings

Claude Opus 5.5 leads modelgrep’s November 2026 ranking of vision-capable AI models with an intelligence score of 59.1. GPT-6 Astra follows at 56.9, ahead of GPT-6.1 Sol at 54.6. The leaderboard covers 25 multimodal models that accept image input for tasks involving images, documents and charts, ordering them by intelligence rather than speed or price.

Claude Sonnet 5.5 sits just behind the leaders with a score of 54.5, alongside listed throughput of 135 t/s and pricing of $2.00/M. GPT-6.1 Terra scores 51.3, while Google’s Gemini 4 Argon reaches 50.4. These entries offer additional comparisons across capability, cost and response speed.

Open models also feature in the ranking. Kimi K3.1 Pro scores 46.5, ahead of other open entries including DeepSeek V4.1 Flash 1017 and Muse Spark Open. Model listings identify capabilities such as reasoning, tool use and JSON support alongside intelligence scores.