Leaderboards

Model leaderboards

Independent Artificial Analysis evaluations for listed models in our model market.

Data: Artificial Analysis · updated: 2026-10-03 · listed market models with matching evaluations only

Intelligence across independent evaluations — higher is better. · Intelligence Index v4.3.2

Numbers rank the models included in our market. Names retain the evaluated version and settings; trial results depend on your chosen parameters. Elo intervals show 95% confidence. Scores from different boards are not comparable.

Intelligence Index vs. Cost per Task

Artificial Analysis Intelligence Index · weighted average cost (USD, list price) per Intelligence Index task

Most attractive quadrantPareto frontierAnthropicOpenAIGrokMoonshotGoogleDeepSeek

Top-left is better: higher score, lower price. Click a point to open the model; the green area holds models below median price and above median score. Hover points whose label didn't fit.

Claude Opus 5.5 (max with fallback)Anthropic58$1.68 in · $8.40 out 2

Claude Sonnet 5.5 (max with fallback)Anthropic56Launch reference · $0.84 input / $4.2 output · per 1M tokens 3

GPT-6 Astra (max)OpenAI53$2.80 in · $14.00 out 4

GPT-6.1 Sol (max)OpenAI52Launch reference · $0.56 input / $2.8 output · per 1M tokens 5

Grok 4.7 (xhigh)Grok46$0.84 in · $2.52 out 6

Kimi K3 (max)Moonshot44$2.95 in · $14.75 out 7

Gemini 3.8 Flash (high)Google41$0.30 in · $1.50 out 8

DeepSeek V4.1 Flash (max)DeepSeek39Launch reference · $0.1125 input / $0.45 output · per 1M tokens 9

GPT-6 Luna (max)OpenAI38Launch reference · $0.028 input / $0.14 output · per 1M tokens