LLM pricing 2026

Every major LLM, side by side — input and output cost per million tokens, context window and modality. Click any column header to sort. Prices reflect the provider's standard API tier; volume, batch and cached input discounts are not included.

Modality
Meta (hosted)Llama 4 Scout$0.11$0.341MText, Image
OpenAIGPT-5 mini$0.25$2.00400kText, Image
Meta (hosted)Llama 4 Maverick$0.27$0.851MText, Image
DeepSeekDeepSeek V3$0.27$1.10128kText
GoogleGemini 2.5 Flash$0.30$2.501MText, Image, Audio, Video
MistralCodestral$0.30$0.90256kText (code)
AlibabaQwen 3 Max$0.40$1.60256kText, Image
DeepSeekDeepSeek-R1$0.55$2.19128kText
AnthropicClaude Haiku 4$0.80$4.00200kText, Image
GoogleGemini 2.5 Pro$1.25$10.001MText, Image, Audio, Video
MistralMistral Large 2$2.00$6.00128kText
CohereCommand A$2.50$10.00256kText
AnthropicClaude Sonnet 4$3.00$15.00200kText, Image
OpenAIGPT-5$5.00$15.00400kText, Image
xAIGrok 4$5.00$15.00256kText, Image
AnthropicClaude Opus 4$15.00$75.00200kText, Image

Prices are USD per million tokens, sourced from each provider's public pricing page and rounded. Subject to change — verify on the provider's site before contracting.

Cheapest frontier
Llama 4 Scout
$0.11 in / $0.34 out
Most expensive
Claude Opus 4
$15 in / $75 out
Longest context
Gemini 2.5 / Llama 4
1,000,000 tokens

FAQ

What is a token?

A sub-word unit of text. Roughly 1 token ≈ 0.75 English words. Pricing is quoted per million tokens (Mtok).

Why is output more expensive than input?

Generation is sequential and uses more GPU time per token, while input can be processed in parallel. Output is typically 3–10× input price.

Which LLM is cheapest?

For frontier-class quality, Llama 4 Scout and Gemini 2.5 Flash are the cheapest mainstream options at ~$0.10–$0.30 per million input tokens.

Which LLM is most expensive?

Claude Opus 4 leads at $15 in / $75 out per million tokens, followed by GPT-5 and Grok 4 at ~$5 in / $15 out.

Go deeper