LLM token pricing, compared.

Every provider publishes token pricing a little differently, which makes apples-to-apples comparison harder than it should be. This page normalizes it: one table, input and output rates per million tokens, cheapest to most expensive. To price your own prompt against any of these, use the calculator.

LLM token pricing table (per million tokens)

Model	Input	Output
Gemini 2.5 Flash-Lite	$0.10	$0.40
GPT-4.1 Nano	$0.10	$0.40
DeepSeek V3.2	$0.14	$0.28
Claude Haiku 4.5	$1.00	$5.00
Gemini 3.5 Flash	$1.50	$9.00
Gemini 3.1 Pro	$2.00	$12.00
Claude Opus 4.8	$5.00	$25.00
Claude Opus 4.8 Fast / Fable 5	$10.00	$50.00

See every model, including the ones not listed here, on the all-models page.

How to read AI token pricing

Two rules make the whole table make sense:

• Prices are per million tokens. A token is about 0.75 words of English, so a million tokens is roughly 750,000 words. See AI token cost explained for the token basics.
• Output costs ~5× input. The two numbers are never equal. Whether input or output dominates your bill depends on your workload — which is why the "cheapest" model changes with the job.

Cheapest token pricing depends on your input/output mix

The raw floor is Gemini 2.5 Flash-Lite and GPT-4.1 Nano at $0.10/$0.40. But if your workload is output-heavy (content generation, long answers), DeepSeek V3.2's $0.28 output rate wins. If it is input-heavy (RAG, document Q&A), the $0.10-input models win. We break this down by workload in the cheapest LLM API by workload.

The pricing that isn't on the table: discounts

Headline token pricing is the starting point, not the final bill. Two discounts change the math:

• Prompt caching — cached input billed at ~10% of the input rate. Big for repeated prompts. See the caching cookbook.
• Batch API — flat 50% off for jobs that can wait 24 hours.

FAQ

What is LLM token pricing?

The per-token rate a provider charges, quoted per million tokens, with separate rates for input and output. Output is usually about 5× the input rate.

Which LLM has the cheapest token pricing in 2026?

Gemini 2.5 Flash-Lite and GPT-4.1 Nano at $0.10/$0.40, with DeepSeek V3.2 ($0.14/$0.28) cheapest for output-heavy work.

Why do providers price input and output separately?

Generating output is more compute-intensive than reading input, so output is priced about 5× higher. Your input/output ratio decides which model is cheapest for you.

How do I compare token pricing for my workload?

Enter your typical input and output token counts into the gpt-cost.com calculator and switch between models — it applies each provider's pricing and shows per-call and monthly cost.

Prices synced from each vendor's official pricing page, July 2026. For exact billing, use each provider's official tokenizer.