LLM Leaderboard - AI Models Ranked by Benchmarks & Price

Based on our live composite scoring across benchmarks, pricing, speed, and capabilities, GPT-5.4 Image 2 currently leads the AI model leaderboard with a score of 92. Rankings are updated hourly using real-time data from 56+ providers.

We use a composite scoring system (0-100) that combines: benchmark performance (90%) from MMLU, GPQA, HumanEval, SWE-bench, and 15+ standardized evaluations, with capabilities and context window as tiebreakers (10%). Data is sourced from provider catalogs, benchmark feeds, and pricing sources, with page freshness shown from the underlying cache.

We track 367 AI models across 56 providers including OpenAI, Anthropic, Google, DeepSeek, Meta, and Mistral. This includes 341 coding models, 15 image generation models, and 10 video generation models.

Our coding category ranks 341 models by their performance on coding benchmarks like SWE-bench, HumanEval, and real-world coding tasks. The top coding models are recalculated as benchmark and pricing sources refresh. Visit our coding leaderboard for the current top 10.