Live AI pricing

AI Token Prices

Current input & output token prices for OpenAI, Claude, Gemini and every major LLM provider — compared in one place, in USD.

Prices per 1,000,000 tokens · Data as of 17 June 2026

Every major AI provider bills by the token — roughly three-quarters of a word — and charges separately for input tokens (your prompt) and output tokens (the model’s reply), with output almost always the pricier of the two. Prices vary by more than 100x between the cheapest lightweight models and the top-tier reasoning models, so the model you pick is usually the single biggest lever on your AI bill.

The table below lists current published prices per 1,000,000 tokens across every provider SpendLil tracks. If you’re trying to cut spend, see model routing and AI spend governance.

AI model token prices per 1 million tokens, in USD. Data as of 17 June 2026.
Anthropicclaude-haiku-4-5US$1.00US$5.00
Anthropicclaude-opus-4-5US$5.00US$25.00
Anthropicclaude-sonnet-4-5US$3.00US$15.00
Coherecommand-rUS$0.15US$0.60
Coherecommand-r-plusUS$2.50US$10.00
DeepSeekdeepseek-chatUS$0.28US$0.42
DeepSeekdeepseek-reasonerUS$0.28US$0.42
Googlegemini-2.0-flashUS$0.10US$0.40
Googlegemini-2.5-flashUS$0.30US$2.50
Googlegemini-2.5-proUS$1.25US$10.00
OpenAIgpt-4-turboUS$10.00US$30.00
OpenAIgpt-4oUS$2.50US$10.00
OpenAIgpt-4o-miniUS$0.15US$0.60
OpenAIo1US$15.00US$60.00
OpenAIo3-miniUS$1.10US$4.40

Prices in USD, sourced from public provider pricing and updated daily. Showing 15 models.

How this data is sourced

Prices are pulled from public provider pricing (via the open-source LiteLLM dataset) and refreshed daily, so this page reflects current list prices rather than a one-off snapshot. Model names are shown in canonical form — date-suffixed variants like gpt-4o-mini-2024-07-18 are collapsed to gpt-4o-mini. GBP figures use the daily Bank of England USD/GBP reference rate. These are list prices and exclude volume discounts, batch pricing and cached-input rates; your actual spend depends on real usage and any negotiated terms — which is exactly what SpendLil measures, by logging every request that passes through it.

Frequently asked questions

What is a token?
A token is the unit AI models read and bill in — roughly three-quarters of an English word, or about 4 characters. A 1,000-word document is around 1,300 tokens. Providers price per million tokens and charge separately for the tokens you send (input) and the tokens the model generates (output).
Why are input and output tokens priced differently?
Generating text is more computationally expensive than reading it, so output tokens almost always cost more than input tokens — often three to five times as much. A model’s real cost therefore depends heavily on how much text it produces, not just how much you send it.
What is the cheapest LLM API?
The lowest-cost options are lightweight models such as GPT-4o mini, Google Gemini Flash and Claude Haiku, which can be more than 100 times cheaper than top-tier reasoning models. The cheapest capable model for a given task — not the cheapest overall — is what actually minimises spend.
How often do AI token prices change?
Provider list prices change infrequently — typically every few months — but cuts can be sudden and large. This page refreshes daily from public pricing, so it reflects the current rate rather than a stale snapshot.
Are these prices in USD or GBP?
Providers publish prices in US dollars, shown in USD by default. Use the USD/GBP toggle to convert at the latest Bank of England daily reference rate.
Do these prices include discounts or batch pricing?
No — these are standard published list prices. Many providers offer cheaper batch processing, cached-input pricing or volume discounts, so your effective cost can be lower depending on how you use the API.

Further reading

Prices change under you. Know what you’re actually spending.

SpendLil sits between your apps and your AI providers and tracks every request — by model, by key, by penny. Add one header. That’s it.

Start tracking free