AI Models

AI Models Compared

Every major frontier model - pricing, context window, benchmark scores, and what each one is actually best at. Updated May 2026.

Pricing at a glance

ModelProviderContextInput / 1MOutput / 1MType
Claude Sonnet 4.6Anthropic200K$3.00$15.00Closed
Claude Opus 4.7Anthropic200K$15.00$75.00Closed
Claude Haiku 4.5Anthropic200K$0.80$4.00Closed
GPT-5OpenAI128K$10.00$30.00Closed
GPT-4oOpenAI128K$2.50$10.00Closed
o1OpenAI200K$15.00$60.00Closed
Gemini 2.5 ProGoogle DeepMind1M$1.25$10.00Closed
Gemini 2.5 FlashGoogle DeepMind1M$0.075$0.30Closed
Amazon Nova ProAmazon Web Services300K$0.80$3.20Closed
DeepSeek V3DeepSeek128K$0.27$1.10Open
Llama 4Meta10MFreeFreeOpen
Qwen 3Alibaba (Qwen Team)128KFreeFreeOpen
Mistral LargeMistral AI128K$3.00$9.00Closed
Gemma 3Google DeepMind128KFreeFreeOpen
Command R+Cohere128K$2.50$10.00Closed

Prices in USD. Open-source models are free to self-host; API pricing varies by provider.

Closed-source models

Claude Sonnet 4.6

200K ctx

Anthropic

Anthropic's best balance of speed, intelligence, and cost for production workloads

Production API integrationsCoding assistants and IDEsDocument analysis
Full details →

Claude Opus 4.7

200K ctx

Anthropic

Anthropic's most capable model for tasks that demand deep reasoning and precision

Complex research tasksHigh-stakes code generationLong-document analysis
Full details →

Claude Haiku 4.5

200K ctx

Anthropic

Anthropic's fastest and most cost-efficient model for high-volume, lightweight tasks

High-volume API pipelinesReal-time classificationForm and document extraction
Full details →

GPT-5

128K ctx

OpenAI

OpenAI's most capable general-purpose model with strong multimodal and reasoning abilities

General-purpose API integrationsMultimodal applicationsComplex reasoning tasks
Full details →

GPT-4o

128K ctx

OpenAI

OpenAI's multimodal workhorse - fast, affordable, and widely integrated

Multimodal appsHigh-volume production useChatbots and assistants
Full details →

o1

200K ctx

OpenAI

OpenAI's reasoning model that thinks before it answers - best for hard science and math

Math and science problemsCompetitive programmingComplex multi-step reasoning
Full details →

Gemini 2.5 Pro

1M ctx

Google DeepMind

Google's most capable model with a 1M token context and top science benchmark scores

Very long document analysisVideo and multimodal understandingScientific research tasks
Full details →

Gemini 2.5 Flash

1M ctx

Google DeepMind

Google's fastest and cheapest model with a 1M context - hard to beat on price/performance

High-volume, long-context tasksCost-sensitive production workloadsDocument and media summarization
Full details →

Amazon Nova Pro

300K ctx

Amazon Web Services

AWS's multimodal frontier model - natively integrated with Bedrock and the AWS ecosystem

AWS-native production workloadsEnterprise deployments requiring governanceVideo and multimodal analysis
Full details →

Mistral Large

128K ctx

Mistral AI

Europe's leading frontier model - strong on code, multilingual tasks, and function calling

Multilingual European deploymentsTool-calling and agentic pipelinesCode generation and review
Full details →

Command R+

128K ctx

Cohere

Cohere's enterprise model purpose-built for RAG and production tool-calling pipelines

Enterprise RAG applicationsKnowledge base Q&A with citationsMulti-step agentic workflows
Full details →

Open-source models

Free to download, self-host, and fine-tune.

Benchmark scores sourced from official provider release posts. Prices subject to change - check provider pricing pages for current rates.