Cached input tokens are usually discounted heavily. These pages show the effective input price and monthly cost at 0–95% cache hit rates.
DeepSeek · Open-Weights
OpenAI · Cost-Efficient
Anthropic · Cost-Efficient
DeepSeek · Open-Weights
Google · Cost-Efficient
OpenAI · Cost-Efficient
Google · Cost-Efficient
OpenAI · Cost-Efficient
OpenAI · Flagship
Anthropic · Flagship
Anthropic · Flagship
Google · Reasoning
Google · Reasoning
xAI · Cost-Efficient
Anthropic · Reasoning
Alibaba · Open-Weights
OpenAI · Flagship
xAI · Flagship
Meta · Open-Weights
OpenAI · Reasoning
Mistral · Flagship
OpenAI · Flagship
Cohere · Flagship