Home › AI model prices

AI model API prices

What the major frontier models cost through their official APIs, in US dollars per 1 million tokens, plus how much text each one can take in at once. 10 models, each checked on the provider's own pricing page.

Standard tier, shortest-context rate. Batch, cached and long-prompt rates differ; see each model's notes and source. Last checked Oct 14, 2026 by the AgentPlays editors.

Input and output price per 1M tokens

Input Output · sorted by output price, highest first

Claude Fable 5.1Anthropic · 1M context

In$10

Out$50

GPT-6 AstraOpenAI · 1.05M context

In$10

Out$50

Claude Opus 5.5Anthropic · 1M context

In$4

Out$20

Muse WonderMeta · 1M context

In$4

Out$20

Claude Sonnet 5.5Anthropic · 1M context

In$2

Out$10

GPT-6.1 SolOpenAI · 1.05M context

In$2

Out$10

Gemini 4 ArgonGoogle · 1M context

In$2

Out$10

Intro price until Nov 30, 2026, then $4 / $20

Mistral Medium 3.5Mistral · 256K context

In$1.50

Out$7.50

Grok 4.7xAI · 500K context

In$2

Out$6

DeepSeek V4 ProDeepSeek · 1M context

In$1.32

Out$3.96

Full comparison

Model	Provider	Input / 1M	Output / 1M	Context window	Notes	Source
Claude Fable 5.1 claude-fable-5-1	Anthropic	$10	$50	1M	Same rate across the full 1M context.	Official pricing ↗ · Model page ↗ docs.anthropic.com Checked Oct 14, 2026
GPT-6 Astra gpt-6-astra	OpenAI	$10	$50	1.05M (922K max input)	Prompts over 272K input tokens cost 2x input and 1.5x output for the whole request.	Official pricing ↗ · Model page ↗ platform.openai.com Checked Oct 14, 2026
Claude Opus 5.5 claude-opus-5-5	Anthropic	$4	$20	1M	Same rate across the full 1M context.	Official pricing ↗ · Model page ↗ docs.anthropic.com Checked Oct 14, 2026
Muse Wonder muse-wonder-1.0	Meta	$4	$20	1M	Muse Wonder Full, the higher-end variant, is not sold through the API. It comes only with Muse Code Power Usage ($50/month) or Maximum Usage ($200/month) plans, per Meta's Muse Wonder page. Source ↗	Official pricing ↗ · Plan prices ↗ dev.meta.ai Checked Oct 14, 2026
Claude Sonnet 5.5 claude-sonnet-5-5	Anthropic	$2	$10	1M	Same rate across the full 1M context.	Official pricing ↗ · Model page ↗ docs.anthropic.com Checked Oct 14, 2026
GPT-6.1 Sol gpt-6.1-sol	OpenAI	$2	$10	1.05M (922K max input)	Prompts over 272K input tokens cost 2x input and 1.5x output for the whole request.	Official pricing ↗ · Model page ↗ platform.openai.com Checked Oct 14, 2026
Gemini 4 Argon gemini-4-argon	Google	$2	$10	1M (input + output combined)	Introductory price through Nov 30, 2026; $4 in / $20 out from Dec 1. The 1M limit is one combined budget for input and output tokens. Paid tier only.	Official pricing ↗ · Model page ↗ ai.google.dev Checked Oct 14, 2026
Mistral Medium 3.5 mistral-medium-3-5	Mistral	$1.50	$7.50	256K	Open weights (Modified MIT license); price is for Mistral's API.	Official pricing ↗ docs.mistral.ai Checked Oct 14, 2026
Grok 4.7 grok-4.7	xAI	$2	$6	500K	Prompts of 200K tokens or more are billed at $4 in / $12 out for the whole request.	Official pricing ↗ · Model page ↗ docs.x.ai Checked Oct 14, 2026
DeepSeek V4 Pro deepseek-v4-pro	DeepSeek	$1.32	$3.96	1M	Peak-hour price shown. Off-peak is half ($0.66 in / $1.98 out); peak is 01:00-04:00 and 06:00-10:00 UTC on weekdays.	Official pricing ↗ api-docs.deepseek.com Checked Oct 14, 2026

How to read this

• Input is what you pay for the text, files and images you send. Output is what the model writes back, including reasoning tokens on thinking models.
• Context window is the most tokens one request can hold. 1M tokens is very roughly 700,000 English words. Some providers count input and output against one shared limit.
• Lowest output price here: DeepSeek V4 Pro at $3.96. Largest listed window: GPT-6.1 Sol and GPT-6 Astra at 1.05M (922K max input).
• Price is not quality. A cheaper model can need more retries; test on your own task before switching.

How we check prices

Every number comes from the provider's official pricing or model page, linked in the Source column with the date we checked it. Models whose price we could not confirm on an official page are left out rather than estimated. Prices change often, so confirm on the source before you budget. Spot an error? See our corrections policy.