Home › AI models explained › Claude vs ChatGPT vs Gemini

Claude vs ChatGPT vs Gemini: API price, context and limits compared

Anthropic, OpenAI and Google each sell a ladder of models from flagship to cheapest. This page lines the tiers up using each provider's own model and pricing pages: price, context window, output limit, caching and batch. It compares the models as sold through APIs, not the chat apps. Compare every model in the price comparison or run your numbers in the cost calculator. Section hub: AI models explained.

Quick answerCompared through their APIs, Claude's current models list 1M-token context windows, GPT-6 1.05M and Gemini 4 1M. At the cheapest tier, GPT-6 Luna and Claude Haiku 5.5 are both $0.10 input / $0.50 output per million tokens; at the top, Claude Fable 5.2 is $8 / $40 and GPT-6 Astra $10 / $50.

Prices last verified against official provider pages on November 13, 2026. How we verify prices. Written by the Chelure Enterprises Editorial Team.

The tiers side by side (USD per million tokens)

Tier	Anthropic Claude	OpenAI GPT-6	Google Gemini 4
Flagship	Claude Fable 5.2: $8 in / $40 out	GPT-6 Astra: $10 in / $50 out	Gemini 4 Carbon: $2 in / $10 out
Strong all-rounder	Claude Opus 5.5: $4 in / $20 out	GPT-6.1 Sol: $2 in / $10 out	Gemini 4 Argon: $2 in / $10 out
Fast and efficient	Claude Sonnet 5.5: $2 in / $10 out	GPT-6.1 Terra: $1 in / $5 out	Gemini 4 Hydrogen: $1.5 in / $7.5 out
Smallest and cheapest	Claude Haiku 5.5: $0.1 in / $0.5 out	GPT-6 Luna: $0.1 in / $0.5 out	No Gemini 4 model at this tier in our table

Tier placement is ours, based on each provider's own positioning (flagship for the most demanding work, down to the most efficient model for high-volume tasks). Gemini 4 Carbon and Argon double to $4 / $20 on December 1, 2026; see Gemini 4 pricing.

Specifications

	Claude (current models)	GPT-6 family	Gemini 4 family
Context window	1M tokens	1.05M tokens	1M tokens
Max output	128K tokens	128K tokens	65K / 1M tokens (as listed)
Knowledge cutoff (as listed)	June 2026	April 30 to June 1, 2026 depending on model	Not shown on Google's models page
Prompt caching	Explicit cache_control; reads 0.1x base input on most models, 0.05x on Opus 5.5 and Sonnet 5.5	Automatic; reads 0.1x input, 0.05x for GPT-6.1 Sol and Terra	Implicit, automatic for Gemini 2.5 and newer
Batch discount	50% off	50% off	50% off
Long-prompt surcharge	Only Haiku 5.5, over 100K tokens	Above 272K input tokens	Gemini 3.1 Pro above 200K (Gemini 4 not stated on the page we read)

What a request costs

For 1,000 chat turns of 1,500 input and 400 output tokens (uncached): Claude Sonnet 5.5 $7.00, GPT-6.1 Terra $3.50, Gemini 4 Hydrogen $5.25; one tier up, Claude Opus 5.5 $14.00, GPT-6.1 Sol $7.00, Gemini 4 Argon $7.00. See the individual guides: Claude Sonnet, GPT-6 and Gemini 4.

How to choose

Start from your workload, not a brand: how long are your prompts, how long are the answers, and how many requests a day?
Test on your own data. Public benchmarks are not your task.
Watch the footnotes: scheduled price changes, long-prompt tiers and cache minimums can change the winner.

Frequently asked questions

What is the difference between Claude, ChatGPT and Gemini?

They are three families of language models from Anthropic, OpenAI and Google. ChatGPT, the Claude app and the Gemini app are products built on them; this page compares the models as sold through their APIs, using each provider's own model and pricing pages.

Which has the biggest context window?

On the providers' own pages, the current Claude models list 1M tokens, GPT-6 models list 1.05M tokens and the Gemini 4 models list 1M tokens, so the differences are small. See the context window table.

Which is cheapest?

It depends on the tier you compare. At the smallest tier, GPT-6 Luna and Claude Haiku 5.5 both list $0.10 per million input tokens and $0.50 output; at the top tiers Claude Fable 5.2 is $8 / $40, GPT-6 Astra $10 / $50 and Gemini 4 Carbon $2 / $10 now, rising to $4 / $20 on December 1, 2026.

Do you compare subscription plans?

No. Chat subscriptions change often and we have not verified their prices and limits on the providers' own pages, so this page compares API prices and specifications only.

Which model is best?

We do not rank quality: benchmarks change quickly and depend on the task. Test two or three models on your own data and compare accuracy against cost with our calculators.

Sources

Primary sources opened on November 13, 2026. Prices come from our verified price table (see the methodology).