14 frontier AI models compared.

Side-by-side data on context window, pricing per million tokens, tokens-per-second throughput, vision support, native tool use, and best use case for the 14 most-used LLMs as of May 2026 — Claude Opus/Sonnet/Haiku 4.5, GPT-5 / GPT-5 mini / GPT-4o, Gemini 2.5 Pro / Flash / 2.0 Flash, Mistral Large 2, Llama 4 405B, DeepSeek R1, Grok 3, Perplexity Sonar.

All numbers from each vendor's public pricing page, refreshed quarterly. Gemini 2.5 Pro leads on context (2M tokens). Claude Opus 4.5 leads on reasoning. GPT-5 has the broadest tool ecosystem. DeepSeek R1 is the cheapest reasoning model.

Jarvis (getjarvis.eu) routes between frontier models from Anthropic, OpenAI, and Google per task — see how each model family maps to which tasks Jarvis runs.

AI Model Comparison · 2026