FindNChoose is reader-supported. Some links may earn us a commission at no cost to you; it never changes our verdicts. How we make money · How we test
Prices per million tokens
| Model | Provider | Input / 1M tokens | Output / 1M tokens | Cached input | Notes |
|---|---|---|---|---|---|
| GPT-6 Astra | OpenAI | $10.00 | $50.00 | $1.000 | Short-context standard price; long context costs more. |
| GPT-6.1 Sol | OpenAI | $2.00 | $10.00 | $0.100 | Short-context standard price. |
| GPT-6.1 Terra | OpenAI | $1.00 | $5.00 | $0.050 | Short-context standard price. |
| GPT-6 Luna | OpenAI | $0.100 | $0.500 | $0.010 | Short-context standard price. |
| Claude Fable 5.1 | Anthropic | $10.00 | $50.00 | $0.250 | |
| Claude Opus 5.5 | Anthropic | $4.00 | $20.00 | $0.200 | |
| Claude Sonnet 5.5 | Anthropic | $2.00 | $10.00 | $0.100 | |
| Claude Haiku 5.5 | Anthropic | $0.100 | $0.500 | $0.010 | Prompts up to 100,000 tokens; longer prompts cost $0.50 in / $2.50 out. |
| Gemini 4 Argon | $2.00 | $10.00 | – | Price through Nov 30, 2026; Google lists $4.00 in / $20.00 out from Dec 1, 2026. Output includes thinking tokens. | |
| Gemini 3.8 Flash | $0.750 | $3.75 | – | Price through Dec 31, 2026; Google lists $1.50 in / $7.50 out from Jan 1, 2027. | |
| Gemini 3.1 Flash-Lite | $0.250 | $1.50 | – | Text, image and video input; audio input costs more. | |
| DeepSeek Flash (V4.1) | DeepSeek | $0.300 | $1.20 | $0.006 | Peak-hour price (cache miss). Off-peak is half. |
| DeepSeek V4 Pro | DeepSeek | $1.32 | $3.96 | $0.044 | Peak-hour price (cache miss). Off-peak is half. |
Official list prices checked Oct 30, 2026, before tax. Sources: platform.openai.com, docs.claude.com, ai.google.dev, api-docs.deepseek.com.
Prices are standard (non-batch) list prices for short prompts, as published by each provider. “Cached input” is the price for input that matches a recent prompt prefix; it is much cheaper on every provider that lists it.
Price tiers at a glance
| Tier | Models | Input / output per 1M |
|---|---|---|
| Frontier | GPT-6 Astra, Claude Fable 5.1 | $10 / $50 |
| Upper mid | Claude Opus 5.5 | $4 / $20 |
| Mid | GPT-6.1 Sol, Claude Sonnet 5.5, Gemini 4 Argon | $2 / $10 |
| Value | GPT-6.1 Terra, DeepSeek V4 Pro, Gemini 3.8 Flash | $0.75 to $1.32 / $3.75 to $5 |
| Budget | GPT-6 Luna, Claude Haiku 5.5, DeepSeek Flash, Gemini 3.1 Flash-Lite | $0.10 to $0.30 / $0.50 to $1.50 |
A worked example
Say your app makes 1,000 requests a day (30,000 a month), each with 2,000 input tokens and 500 output tokens. That is 60 million input and 15 million output tokens a month. With no caching:
| Model | Monthly cost |
|---|---|
| GPT-6 Luna | $13.50 |
| Claude Haiku 5.5 | $13.50 |
| DeepSeek Flash (peak price) | $36.00 |
| Gemini 3.1 Flash-Lite | $37.50 |
| Gemini 3.8 Flash | $101.25 ($202.50 from Jan 1, 2027) |
| GPT-6.1 Terra | $135.00 |
| DeepSeek V4 Pro (peak price) | $138.60 |
| GPT-6.1 Sol, Claude Sonnet 5.5, Gemini 4 Argon | $270.00 (Gemini 4 Argon $540 from Dec 1, 2026) |
| Claude Opus 5.5 | $540.00 |
| GPT-6 Astra, Claude Fable 5.1 | $1,350.00 |
Calculated from list prices checked October 31, 2026. Try your own numbers in the AI API cost calculator.
The spread is a factor of 100 between the cheapest and most expensive models for the same traffic. For classification, extraction, tagging or short summaries, a budget model is often good enough. Test it before paying for a bigger one.
How caching changes the bill
If most of your prompt is a fixed system prompt or a shared document, prompt caching cuts input cost sharply. In the example above with Claude Sonnet 5.5, serving 80% of input from cache lowers input cost from $120 to $28.80 a month, and the total from $270 to $178.80. Output is unaffected, so caching matters most for long prompts with short answers.
Announced price changes
- Gemini 4 Argon: Google lists $2 in and $10 out through November 30, 2026, then $4 and $20 from December 1, 2026. Google also notes that output pricing includes thinking tokens.
- Gemini 3.8 Flash: $0.75 in and $3.75 out through December 31, 2026, then $1.50 and $7.50 from January 1, 2027.
- DeepSeek: listed prices are peak-hour rates; off-peak usage costs half.
- Claude Haiku 5.5: the listed price covers prompts up to 100,000 tokens; longer prompts cost $0.50 in and $2.50 out.
- OpenAI: listed prices are short-context standard rates; long-context requests on GPT-6 Astra cost more.
Hidden costs to plan for
- Reasoning tokens. Reasoning models think before answering, and those tokens are billed as output. They can be several times longer than the visible answer.
- Conversation history. Chat apps resend previous messages with every request, so input grows with each turn.
- Retries and agents. An agent that takes ten steps makes ten requests.
- Tools and search. Some providers charge separately for built-in web search or other tools.
How to lower your API bill
- Start with the cheapest model that might work, and move up only where quality falls short.
- Route by difficulty: send easy requests to a budget model and hard ones to a bigger one.
- Put fixed instructions at the start of the prompt so they can be cached.
- Cap output length and ask for concise answers.
- Trim or summarize long conversation history.
- Use batch APIs for work that does not need an instant reply.
API vs subscription
A ChatGPT or Claude subscription does not include API access (apart from credits on some top tiers). If you are building a product, you pay for the API separately. If you only want to use AI for your own coding, a subscription is usually cheaper; see Cursor vs Claude Code and ChatGPT vs Claude for coding.
Of the models we track, GPT-6 Luna and Claude Haiku 5.5 list lowest at $0.10 per million input tokens and $0.50 per million output tokens. DeepSeek Flash is $0.30 in and $1.20 out at peak, and half that off-peak. On October 31, 2026, OpenAI listed GPT-6 Astra at $10 in and $50 out per million tokens, GPT-6.1 Sol at $2 and $10, GPT-6.1 Terra at $1 and $5, and GPT-6 Luna at $0.10 and $0.50. Anthropic listed Claude Fable 5.1 at $10 in and $50 out per million tokens, Opus 5.5 at $4 and $20, Sonnet 5.5 at $2 and $10, and Haiku 5.5 at $0.10 and $0.50 for prompts up to 100,000 tokens. Yes, for two models. Google lists Gemini 4 Argon doubling to $4 in and $20 out on December 1, 2026, and Gemini 3.8 Flash doubling to $1.50 and $7.50 on January 1, 2027.Frequently asked questions
Which AI API is the cheapest?
How much does the OpenAI API cost?
How much does the Claude API cost?
Is the Gemini API getting more expensive?
Sources
We run our own WordPress sites on shared hosting and on a VPS, and use AI coding tools every day. We publish what we measured, list every price with the date we checked it and label anything we have not tested ourselves. About us · Editorial policy