How LLM API prices work
Every provider here charges per token, quoted per million tokens. A token is a piece of a word; a page of English is roughly 500 to 700 tokens, but each provider counts a little differently.
- Input is everything you send: your prompt, the system prompt, the chat so far, and any files.
- Output is what the model writes back, including thinking tokens on models that think. It costs several times as much as input.
- Cached input is input the provider read from its cache because you sent the same start recently. It costs a small fraction of the input price.
- Some models charge more once one prompt passes a size line, and batch jobs cost half the standard price at Anthropic, OpenAI, and Google.
Anthropic (Claude) API prices
Haiku 5.5 charges more once a prompt passes 100K tokens. Prompt caching needs cache markers, and a 5 minute cache write costs 1.25 times the input price. Ways to lower the bill are in how to cut your Anthropic API costs.
| Model | Input | Cached input | Output | Notes |
|---|---|---|---|---|
| Claude Opus 5.5 | $4 | $0.20 | $20 | None |
| Claude Sonnet 5.5 | $2 | $0.10 | $10 | None |
| Claude Haiku 5.5 | $0.10 | $0.01 | $0.50 | Over 100K prompt tokens: $0.50 in, $2.50 out. |
| Claude Haiku 4.5 | $1 | $0.10 | $5 | None |
Anthropic notes that Claude 4.7 and later count more tokens for the same text than older models, so compare the cost of real tasks, not only the price per token.
OpenAI (GPT) API prices
Caching is automatic for repeated prompt starts, and cache writes cost nothing extra. Regional processing adds 10 in 100 to the price for models released from March 5, 2026. More in how to cut your OpenAI API costs.
| Model | Input | Cached input | Output | Notes |
|---|---|---|---|---|
| GPT-5.5 | $5 | $0.50 | $30 | Over 272K prompt tokens: $10 in, $45 out. |
| GPT-6.1 Sol | $2 | $0.10 | $10 | Over 272K prompt tokens: $4 in, $15 out. |
| GPT-6.1 Terra | $1 | $0.05 | $5 | Over 272K prompt tokens: $2 in, $7.50 out. |
| GPT-6 Luna | $0.10 | $0.01 | $0.50 | Over 272K prompt tokens: $0.20 in, $0.75 out. |
| GPT-5.4 mini | $0.75 | $0.075 | $4.50 | None |
| GPT-5 mini | $0.25 | $0.025 | $2 | None |
| GPT-5 nano | $0.05 | $0.005 | $0.40 | None |
| GPT-4o mini | $0.15 | $0.075 | $0.60 | Marked deprecated, still served. |
Google (Gemini) API prices
Two Gemini models are on a launch price, shown below with the date it ends and the price after. Output prices include thinking tokens. FUSR.ai does not route to Gemini; these are here so you can compare.
| Model | Input | Cached input | Output | Notes |
|---|---|---|---|---|
| Gemini 4 Argon | $2 | $0.10 | $10 | Launch price until November 30, 2026, then $4 in, $20 out. Not on the free tier. |
| Gemini 3.1 Pro Preview | $2 | $0.20 | $12 | Over 200K prompt tokens: $4 in, $18 out. |
| Gemini 3.8 Flash | $0.75 | $0.075 | $3.75 | Launch price until December 31, 2026, then $1.50 in, $7.50 out. |
| Gemini 3.5 Flash-Lite | $0.30 | $0.03 | $2.50 | None |
| Gemini 3.1 Flash-Lite | $0.25 | $0.025 | $1.50 | None |
| Gemini 2.5 Flash-Lite | $0.10 | $0.01 | $0.40 | None |
xAI (Grok) API prices
Once a prompt reaches 200K tokens, xAI charges the higher price for every token in that request, not only the part over the line.
| Model | Input | Cached input | Output | Notes |
|---|---|---|---|---|
| Grok 4.7 | $2 | $0.50 | $6 | Over 200K prompt tokens: $4 in, $12 out. |
| Grok 4.3 | $1.25 | $0.20 | $2.50 | Over 200K prompt tokens: $2.50 in, $5 out. |
| Grok Build 0.1 | $1 | $0.20 | $2 | 256K window. |
One workload, priced on each model
Prices per token are hard to compare in your head. Here is 20 million input tokens and 4 million output tokens a month, prompts around 4,000 tokens, at list price with no caching or batch. To price your own numbers, use the LLM token cost calculator.
| Model | Input | Output | Per month |
|---|---|---|---|
| Claude Opus 5.5 | $80.00 | $80.00 | $160 |
| Claude Sonnet 5.5 | $40.00 | $40.00 | $80.00 |
| Claude Haiku 5.5 | $2.00 | $2.00 | $4.00 |
| GPT-5.5 | $100 | $120 | $220 |
| GPT-6.1 Terra | $20.00 | $20.00 | $40.00 |
| GPT-5 mini | $5.00 | $8.00 | $13.00 |
| Gemini 4 Argon | $40.00 | $40.00 | $80.00 |
| Gemini 3.8 Flash | $15.00 | $15.00 | $30.00 |
| Grok 4.7 | $40.00 | $24.00 | $64.00 |
| Grok 4.3 | $25.00 | $10.00 | $35.00 |
Gemini rows use the launch price. The cheapest model is only cheaper if it does the job, so try a sample of your own requests on it first.
Prices by task and by provider
- How to cut your OpenAI API costs
- How to cut your Anthropic (Claude) API costs
- Cheapest LLM API for coding, priced per task
- Cheapest LLM API for summarization, worked out
- Cheapest LLM API for classification and extraction
- GPT vs Claude API pricing, model by model
- Claude vs Gemini API pricing, model by model
- GPT vs Gemini API pricing, model by model
- How to reduce your LLM API costs