Claude vs GPT Pricing: Full Breakdown

Head-to-Head: All Claude vs GPT Models

Here is every current Claude and GPT model with its per-token pricing and context window. All prices are per 1 million tokens.

Price Per Word and Per Page

Token pricing can feel abstract. Here is what each model costs when you think in words and pages instead. One token is roughly 0.75 words, so 1 million tokens is about 750,000 words. A standard page is ~500 words (~667 tokens).

Tier-by-Tier Comparison

The fairest comparison matches models of the same tier. Here is how each Claude model stacks up against its GPT counterpart.

Input-Heavy vs Output-Heavy Workloads

Which provider wins depends on what you are doing. Input-heavy workloads (reading documents, analyzing code, processing data) are dominated by input token costs. Output-heavy workloads (writing articles, generating code, producing reports) are dominated by output costs.

Real Dollar Examples

Abstract per-token prices are hard to reason about. Here are three concrete scenarios with real costs on every Claude and GPT model.

The Bottom Line

At the frontier tier, Claude Fable 5.1 vs GPT-6 Astra: the same price, $10 input and $50 output per 1M tokens.

At the flagship tier, Claude Opus 5.5 vs GPT-6 Sol: GPT is cheaper, 2x less on output.

For budget workloads, the cheapest Claude model is Claude Haiku 4.5 at $1 / $5, and the cheapest GPT model is GPT-6 Luna at $0.10 / $0.50. GPT is cheaper, 10x less on output.

Short version: GPT is cheaper at most tiers right now. Claude's case rests on the model itself — pick it when it does the job better, not to save money.