Home / Cheapest AI API
Cheapest AI API, ranked by your job
Pick a job and every model is ranked by what it costs at each company’s list prices. For a short chat reply, Claude Haiku 5.5 and GPT-6 Luna tie as the cheapest. For a long document, GPT-6 Luna is cheapest.
Prices checked October 13, 2026 · Free tiers checked October 20, 2026 · Sources · Price changes
Cheapest first
- 1. Claude Haiku 5.5Anthropic$0.00045
- 1. GPT-6 LunaOpenAI$0.00045
- 3. DeepSeek FlashDeepSeek$0.0012
- 4. Gemini 3.5 Flash-LiteGoogle$0.0019
- 5. Gemini 3.8 FlashGoogle$0.0034
- 6. DeepSeek V4 ProDeepSeek$0.0046
- 7. Grok 4.7xAI$0.007
- 8. Claude Sonnet 5.5Anthropic$0.009
- 8. GPT-6.1 SolOpenAI$0.009
- 8. Gemini 4 ArgonGoogle$0.009
- 11. Claude Opus 5.5Anthropic$0.018
- 12. Claude Fable 5.1Anthropic$0.045
- 12. GPT-6 AstraOpenAI$0.045
The cheapest model from each company
| Company | Model | Input / output per 1M | 1,000 chat replies | 1,000 long documents | 1M tokens each way |
|---|---|---|---|---|---|
| Anthropic | Claude Haiku 5.5 | $0.10 / $0.50 | $0.45 | $80.00long-prompt price | $0.60 |
| OpenAI | GPT-6 Luna | $0.10 / $0.50 | $0.45 | $16.00 | $0.60 |
| Gemini 3.5 Flash-Lite | $0.30 / $2.50 | $1.85 | $50.00 | $2.80 | |
| xAI | Grok 4.7 | $2.00 / $6.00 | $7.00 | $312.00 | $8.00 |
| DeepSeek | DeepSeek Flash | $0.30 / $1.20 | $1.20 | $47.40 | $1.50 |
A model named under a cost is that company’s cheapest model for that job, when it differs. A chat reply is 2,000 tokens in and 500 out; a long document is 150,000 in and 2,000 out.
Free AI APIs
Of the five companies here, only Google lists a free tier on its API pricing page. On Google’s free tier these models cost nothing to use: Gemini 3.8 Flash and Gemini 3.5 Flash-Lite, and the older Gemini 3.6 Flash, Gemini 3.1 Flash-Lite, Gemini 2.5 Pro, Gemini 2.5 Flash and Gemini 2.5 Flash-Lite. Gemini 4 Argon is not on it.
The free tier has lower rate limits, which Google shows for your project in Google AI Studio, and Google says it uses free-tier content to improve its products; on the paid tier it doesn’t. Requests-per-day limits reset at midnight Pacific time, according to Google’s rate limits page.
The other four charge per token. Anthropic says new users get a small amount of free credits to test its API. xAI asks you to load credits before you start, and DeepSeek takes its fees from your topped-up balance. OpenAI’s pricing page lists a price for every GPT model; of its models, only the moderation model is free.
Cheapest is not the whole story: capability for the price
Epoch Capabilities Index (up) against the blended price per 1M tokens, 3 input to 1 output (right, log scale). Dots on the line have no model that is both cheaper and higher scoring. Hover or tap a dot for its numbers.
Show the numbers (16 models)
| # | Model | Company | Score | Blended per 1M | Value line |
|---|---|---|---|---|---|
| 1 | Claude Opus 5.5 | Anthropic | 166.8 | $8.00 | Yes |
| 2 | GPT-6 Astra | OpenAI | 166.4 | $20.00 | |
| 3 | GPT-6.1 Sol | OpenAI | 166.3 | $4.00 | Yes |
| 4 | Claude Fable 5.1 | Anthropic | 164.7 | $20.00 | |
| 5 | Gemini 4 Argon | 164.6 | $4.00 | ||
| 6 | Claude Sonnet 5.5 | Anthropic | 164.4 | $4.00 | |
| 7 | GPT-6 Sol | OpenAI | 162.6 | $4.00 | |
| 8 | Claude Opus 5 | Anthropic | 162.6 | $10.00 | |
| 9 | GPT-5.6 Sol | OpenAI | 161.7 | $8.00 | |
| 10 | GPT-5.6 Terra | OpenAI | 159.4 | $4.50 | |
| 11 | Claude Haiku 5.5 | Anthropic | 159.0 | $0.20 | Yes |
| 12 | Gemini 3.8 Flash | 157.2 | $1.50 | ||
| 13 | GPT-6 Luna | OpenAI | 156.3 | $0.20 | |
| 14 | DeepSeek Flash | DeepSeek | 155.6 | $0.525 | |
| 15 | DeepSeek V4 Pro | DeepSeek | 155.3 | $1.98 | |
| 16 | Grok 4.7 | xAI | 154.4 | $3.00 |
Scores: Epoch AI, Capabilities & benchmarking (Epoch Capabilities Index), CC BY 4.0, checked October 18, 2026. Each score carries a margin of error of about ±3 points, so models a few points apart may be level. Prices: each company’s pricing page, standard tier.
How the ranking works
Each job is priced the same way on every model: tokens in × the input price, plus tokens out × the output price, divided by one million. Prices are each company’s list prices in US dollars, checked October 13, 2026. Models with the same cost share a place.
- Long prompts. Some models charge more once a single prompt passes a set length. With the long document’s 150,000 tokens in, Claude Haiku 5.5 moves to its long-prompt price. The “1M each way” job is spread over normal-sized requests, so it uses standard prices.
- Batch. With “Batch price” ticked, models use their batch price where our list has one: Anthropic, OpenAI and Google, at half the standard price. Other models keep their full price.
- Off-peak. DeepSeek charges half outside its peak hours.
- Not included: cached input (every model here bills it at a lower price), taxes, regional surcharges, tool fees such as web search, and free tiers.
The same text can be a different number of tokens on different models. Anthropic says its Claude 4.7 and later models produce about 30% more tokens for the same text than its earlier tokenizer, so a cheaper price per token doesn’t always mean a cheaper job. The token counter estimates your own text.
Questions
Which AI API is the cheapest?
For a short chat reply, Claude Haiku 5.5 and GPT-6 Luna: $0.45 per 1,000 replies at list prices, a tie. For long documents, GPT-6 Luna, at $16.00 per 1,000. Counting only volume, a million tokens in and a million out costs $0.60 on Claude Haiku 5.5 and GPT-6 Luna. With older models included, GPT-5 nano comes out cheapest for a chat reply, at $0.30 per 1,000.
Is there a free AI API?
Google’s Gemini API has a free tier for Gemini 3.8 Flash and Gemini 3.5 Flash-Lite and several older models, with lower limits, and Google uses free-tier content to improve its products. Anthropic gives new users a small amount of free credits to try its API. Otherwise you pay per token.
Does the batch API make it cheaper?
Yes, for work that can wait: Anthropic, OpenAI and Google list batch prices at half their standard prices. Tick “Batch price” above to see the order change.
Is the cheapest model good enough?
Price says nothing about quality. For independent benchmark rankings next to each model’s price, see Best AI models, and try a cheap model on your own task before you commit.
Why does a long document cost more on some models?
Some companies charge a higher price per token once one prompt passes a set length, for example over 100,000 tokens on Claude Haiku 5.5 and over 272,000 on OpenAI’s GPT-6 models. The cost calculator switches to those prices for you.
Sources
- Anthropic’s pricing page (list prices, checked October 13, 2026; our Claude API pricing page has the full table)
- OpenAI’s pricing page (list prices, checked October 13, 2026; our OpenAI API pricing page has the full table)
- Google’s pricing page (list prices, checked October 13, 2026; our Gemini API pricing page has the full table)
- xAI’s pricing page (list prices, checked October 13, 2026; our Grok API pricing page has the full table)
- DeepSeek’s pricing page (list prices, checked October 13, 2026; our DeepSeek API pricing page has the full table)
- Google’s Gemini API pricing (free tier, checked October 20, 2026) and rate limits
- Anthropic’s pricing FAQ (free credits for new users; tokenizer), xAI’s quickstart, DeepSeek’s deduction rules and OpenAI’s pricing page, read October 20, 2026