Models
LLM API pricing, by provider
A rate card is the start of the answer, not the answer. Each page pairs the published prices with what the model actually costs on three real workloads — tokenizer calibration, caching, and retries included. Computed from pricing data verified 2026-07-11.
OpenAI
GPT-5 API Pricing Explained
Rate card for the GPT-5.4 line, plus the three things that decide your bill: cached input at 10%, no cache-write premium, and a 50% batch tier.
Anthropic
Claude API Cost Explained
Sonnet 4.6 and Haiku 4.5 rates, plus the two Anthropic-specific line items that move your bill more than the sticker price does.
Google
Gemini 2.5 Flash Pricing
The cheapest credible frontier-adjacent option in the catalog — and the only one pairing a 1M-token window with a tokenizer that bills slightly under baseline.
DeepSeek
DeepSeek V4 Flash Pricing (and the Data-Policy Caveat)
An order of magnitude below everything else, with a 98% cache discount — and a data-residency question you have to answer before the price matters.
Weighing two of them against each other? The comparison pages run head-to-heads on the same workloads.