AiGanakLLM API pricing, compared: OpenAI, Anthropic and Google
Current list prices per million tokens for every major model, and a calculator that turns your own traffic into a monthly cost for each one. Prices are copied from each provider's pricing page and dated.
Current list prices per million tokens for GPT-6 Astra, GPT-6.1 Sol, GPT-6 Luna, Claude Fable 5.1, Opus 5.5, Sonnet 5.5, Haiku 5.5 and Gemini models, with cached and batch prices, announced price changes and a free cost calculator. Prices are dated and sourced from each provider.
What would your traffic cost on each model?
Enter your monthly volume and the size of a typical request. Start from a preset if you are not sure.
| Model | Monthly cost | Per 1,000 requests |
|---|
Estimates use list prices. Not included: Anthropic cache-write charges, Google cache storage fees, taxes, discounts and tools such as web search. Quality differs between models, so test on your own examples before switching.
Get a sample cost report for your workload
We email you a report of the numbers above: your workload priced on every current model, what your current model costs, and what a full report would add. Need something specific, such as a comparison for your own mix of models, regions or volumes? Tell us and we will quote a custom report.
Current list prices
Standard pay-as-you-go tier, text tokens. "Cached" is the price of input the provider has seen recently: OpenAI cached input, Anthropic cache reads and Gemini context caching.
OpenAI source ↗
| Model | Input | Cached input | Output | Batch input | Batch output | Notes |
|---|---|---|---|---|---|---|
GPT-6 Astragpt-6-astra | $10 | $1 | $50 | $5 | $25 | Flagship. Cache write $12.50. Prompts over 272K input tokens cost 2x input and cache rates and 1.5x output for the whole request. |
GPT-6.1 Solgpt-6.1-sol | $2 | $0.1 | $10 | $1 | $5 | Mid |
GPT-6 Lunagpt-6-luna | $0.1 | $0.01 | $0.5 | $0.05 | $0.25 | Small |
Anthropic source ↗
| Model | Input | Cached input | Output | Batch input | Batch output | Notes |
|---|---|---|---|---|---|---|
Claude Fable 5.1claude-fable-5-1 | $10 | $0.25 | $50 | $5 | $25 | Flagship. Cache write $12.50 |
Claude Opus 5.5claude-opus-5-5 | $4 | $0.2 | $20 | $2 | $10 | Flagship. Cache write $5 |
Claude Sonnet 5.5claude-sonnet-5-5 | $2 | $0.1 | $10 | $1 | $5 | Mid. Cache write $2.50 |
Claude Haiku 5.5claude-haiku-5-5 | $0.1 | $0.01 | $0.5 | $0.05 | $0.25 | Small. Cache write $0.125. Requests whose prompt is over 100,000 tokens (including cache reads and writes) pay higher rates. |
Claude Haiku 4.5claude-haiku-4-5 | $1 | $0.1 | $5 | $0.5 | $2.50 | Small. Previous generation. Cache write $1.25 |
Google source ↗
| Model | Input | Cached input | Output | Batch input | Batch output | Notes |
|---|---|---|---|---|---|---|
Gemini 4 Argongemini-4-argon | $2 | $0.1 | $10 | $1 | $5 | Flagship. Rises to $4 / $20 on 1 December 2026 |
Gemini 3.1 Pro (preview)gemini-3.1-pro-preview | $2 | $0.2 | $12 | $1 | $6 | Flagship. Preview. Prompts over 200K tokens pay higher rates. |
Gemini 3.8 Flashgemini-3.8-flash | $0.75 | $0.075 | $3.75 | $0.375 | $1.88 | Mid. Rises to $1.50 / $7.50 on 1 January 2027 |
Gemini 3.5 Flash-Litegemini-3.5-flash-lite | $0.3 | $0.03 | $2.50 | $0.15 | $1.25 | Small |
Gemini 3.1 Flash-Litegemini-3.1-flash-lite | $0.25 | $0.025 | $1.50 | $0.125 | $0.75 | Small |
Checked on 24 October 2026. Providers change prices often; confirm on their pages before you commit to a budget. Spotted a change? Tell us.
Prices going up soon
Google has published dated increases for two Gemini models. If you budget on today's rates, plan for these.
Gemini 4 Argon
From 1 December 2026: input $2 → $4, output $10 → $20 per million tokens. That is double the current rate.
Gemini 3.8 Flash
From 1 January 2027: input $0.75 → $1.50, output $3.75 → $7.50 per million tokens. That is double the current rate.
Four things that change your real cost
Input volume often outweighs output price
Output tokens are priced about five times higher, but most applications send far more input: system prompts, retrieved documents and conversation history. Check both sides of your own traffic.
Caching is the cheapest saving
Cached input costs 90% to 97.5% less than normal input on the models above. It only applies when the start of each prompt stays identical, so put instructions first and changing content last.
Batch halves the price
All three providers charge about half for work submitted as a batch and returned within 24 hours. Reports, backfills and nightly jobs rarely need an instant answer.
Long prompts can cost more
Some models charge higher rates above a prompt-length threshold: GPT-6 Astra over 272K tokens, Claude Haiku 5.5 over 100K and Gemini 3.1 Pro over 200K. The calculator applies these automatically.
Want the full method? Read how to reduce LLM API costs: nine levers, in order.
Choose how deep you want to go
Free tools
- Spend by model and monthly run rate
- Every major model priced for your traffic
- Savings estimate from your assumptions
- Private: your file is never uploaded
Sample report
- Your numbers in a branded report you can share
- Your workload priced on every current model
- Savings split by routing, caching and batch
- What the full report would add for you
Full LLM Cost Report
- Request-level analysis of your real calls
- Cacheable prompt prefixes measured, not guessed
- Ranked savings plan with code for your provider
- Which changes need a quality test, and how
- Three-month cost forecast, including announced price rises
LLM Cost & Reliability Audit
- Everything in the full report
- Cheaper models tested on your own approved answers
- Prompt and reliability review
- Walkthrough and implementation plan
LLM API pricing questions
Can't find it? Press ⌘K and ask.
Which LLM API is cheapest?
At list prices checked on 24 October 2026, the cheapest models in our table are OpenAI GPT-6 Luna and Anthropic Claude Haiku 5.5, both at $0.10 per million input tokens and $0.50 per million output tokens (Haiku 5.5 rises above 100,000-token prompts). The cheapest model is only the right choice if it handles your task at the quality you need.
How much does GPT-6 Astra cost?
GPT-6 Astra costs $10 per million input tokens, $1 per million cached input tokens and $50 per million output tokens on the standard tier. Batch and Flex are half price. Prompts over 272,000 input tokens cost twice the input rate and 1.5 times the output rate.
How much does Claude Opus 5.5 cost?
Claude Opus 5.5 costs $4 per million input tokens, $20 per million output tokens, $0.20 per million for cache reads and $5 per million for five-minute cache writes. Message Batches are 50% off.
Are Gemini prices going up?
Yes. Google lists Gemini 4 Argon at $2 input and $10 output per million tokens through 30 November 2026, rising to $4 and $20 from 1 December 2026. Gemini 3.8 Flash and 3.6 Flash double from 1 January 2027.
How is the monthly cost calculated?
Requests per month times the cost of one request, where one request costs uncached input tokens at the input price, cached input tokens at the cached price and output tokens at the output price. Batch uses the batch price and long-prompt rates apply above each model's threshold.
Know what you pay. Now find what you can save.
Upload your OpenAI or Anthropic usage export to the free Bill Analyzer. Your file stays private, and you see your biggest savings in seconds.