LLM API pricing (9 providers, 26 models)

Across 26 models from 9 providers, the lowest output price is Amazon Nova (Bedrock) Amazon Nova Micro at $0.14 per 1M tokens and the highest is OpenAI GPT-6 Astra and Anthropic (Claude) Claude Fable 5.1 at $50. The lowest input price is Amazon Nova (Bedrock) Amazon Nova Micro at $0.035 per 1M tokens. Standard rates as of Oct 18, 2026.

Prices are from each vendor's official pricing page, checked on Oct 18, 2026. This page has no affiliate links.

Data updated Oct 18, 2026Prices checked by handSource: each vendor's official pricing page

Only standard rates shown on the official pricing pages are listed (no batch discounts or priority tiers). Values we could not confirm are shown as "—". Checked Oct 18, 2026.

LLM API prices per 1M tokens, lowest output price first (checked Oct 18, 2026)
Model	Input / 1M tokens	Output / 1M tokens	1M in + 1M out
Amazon Nova MicroAmazon Nova (Bedrock)	$0.035	$0.14	$0.18
Details Provider Amazon Nova (Bedrock) Cached input / 1M tokens — Context — Note US East (N. Virginia), Standard tier. Text only. Source, checked Source Oct 18, 2026
Command R7BCohere	$0.0375	$0.15	$0.19
Details Provider Cohere Cached input / 1M tokens — Context 128K Source, checked Source Oct 18, 2026
Ministral 3 8BMistral AI	$0.15	$0.15	$0.30
Details Provider Mistral AI Cached input / 1M tokens $0.015 Context — Source, checked Source Oct 18, 2026
Amazon Nova LiteAmazon Nova (Bedrock)	$0.06	$0.24	$0.30
Details Provider Amazon Nova (Bedrock) Cached input / 1M tokens — Context — Note US East (N. Virginia), Standard tier. Source, checked Source Oct 18, 2026
GPT-6 LunaOpenAI	$0.10	$0.50	$0.60
Details Provider OpenAI Cached input / 1M tokens $0.01 Context 1.05M Note Rate for input up to 272K. Above 272K: input $0.20, cached $0.02, output $0.75. Cache write $0.125. Source, checked Source Oct 18, 2026
Claude Haiku 5.5Anthropic (Claude)	$0.10	$0.50	$0.60
Details Provider Anthropic (Claude) Cached input / 1M tokens $0.01 Context 1M Note Rate for prompts up to 100,000 tokens. Above 100,000: input $0.50, cached $0.05, output $2.50. Source, checked Source Oct 18, 2026
Mistral Small 4Mistral AI	$0.15	$0.60	$0.75
Details Provider Mistral AI Cached input / 1M tokens $0.015 Context 256K Source, checked Source Oct 18, 2026
Command RCohere	$0.15	$0.60	$0.75
Details Provider Cohere Cached input / 1M tokens — Context 128K Note Listed as "Command R" on the pricing page. The older Command R 03-2024 (existing customers) is input $0.50, output $1.50. Source, checked Source Oct 18, 2026
DeepSeek-V4.1-FlashDeepSeek	$0.30	$1.20	$1.50
Details Provider DeepSeek Cached input / 1M tokens $0.006 Context 1M Note Peak-hours rate. Off-peak is half price (input $0.15, cached $0.003, output $0.6). API model name is deepseek-flash (currently DeepSeek-V4.1-Flash-1017). Source, checked Source Oct 18, 2026
Grok Build 0.1xAI (Grok)	$1	$2	$3
Details Provider xAI (Grok) Cached input / 1M tokens $0.20 Context 256K Note Rate for prompts under 200K tokens. At 200K or more: input $2.00, cached $0.40, output $4.00. Source, checked Source Oct 18, 2026
Mistral Large 4Mistral AI	$0.68	$2.09	$2.77
Details Provider Mistral AI Cached input / 1M tokens $0.07 Context 1M Note The pricing page shows a sale price (regular: input $1.36, cached $0.14, output $4.18). No sale end date is stated. Source, checked Source Oct 18, 2026
Gemini 3.5 Flash-LiteGoogle (Gemini API)	$0.30	$2.50	$2.80
Details Provider Google (Gemini API) Cached input / 1M tokens $0.03 Context 1.05M Note Same input rate for text, image, video and audio. Cache storage $1.00 per 1M tokens per hour. Source, checked Source Oct 18, 2026
Amazon Nova 2 LiteAmazon Nova (Bedrock)	$0.30	$2.50	$2.80
Details Provider Amazon Nova (Bedrock) Cached input / 1M tokens $0.075 Context — Note Global cross-region inference, Standard tier. In-region and Geo inference: input $0.33, output $2.75. Cache reads are 75% off input (per the official page). Source, checked Source Oct 18, 2026
Grok 4.3xAI (Grok)	$1.25	$2.50	$3.75
Details Provider xAI (Grok) Cached input / 1M tokens $0.20 Context 1M Note Rate for prompts under 200K tokens. At 200K or more: input $2.50, cached $0.40, output $5.00. Source, checked Source Oct 18, 2026
Amazon Nova ProAmazon Nova (Bedrock)	$0.80	$3.20	$4
Details Provider Amazon Nova (Bedrock) Cached input / 1M tokens — Context — Note US East (N. Virginia), Standard tier. Varies by region. Source, checked Source Oct 18, 2026
Gemini 3.8 FlashGoogle (Gemini API)	$0.75	$3.75	$4.50
Details Provider Google (Gemini API) Cached input / 1M tokens $0.075 Context 1.05M Note Price valid through 2026-12-31. From 2027-01-01: input $1.50, output $7.50, cached $0.15. Cache storage $0.50 per 1M tokens per hour. Source, checked Source Oct 18, 2026
DeepSeek-V4-ProDeepSeek	$1.32	$3.96	$5.28
Details Provider DeepSeek Cached input / 1M tokens $0.044 Context 1M Note Peak-hours rate (weekdays UTC 01:00-04:00 and 06:00-10:00). Off-peak is half price (input $0.66, cached $0.022, output $1.98). Source, checked Source Oct 18, 2026
Muse Spark 1.3Meta (Model API)	$1.25	$4.25	$5.50
Details Provider Meta (Model API) Cached input / 1M tokens $0.15 Context 1.05M Note Standard mode rate (muse-spark-1.2 and 1.1 share the same price table). Source, checked Source Oct 18, 2026
Grok 4.7xAI (Grok)	$2	$6	$8
Details Provider xAI (Grok) Cached input / 1M tokens $0.50 Context 500K Note Rate for prompts under 200K tokens. At 200K or more, the whole request is billed at input $4.00, cached $1.00, output $12.00. Recommended top model in the official docs. Source, checked Source Oct 18, 2026
Mistral Medium 3.5Mistral AI	$1.50	$7.50	$9
Details Provider Mistral AI Cached input / 1M tokens $0.15 Context 256K Source, checked Source Oct 18, 2026
GPT-6.1 SolOpenAI	$2	$10	$12
Details Provider OpenAI Cached input / 1M tokens $0.10 Context 1.05M Note Rate for input up to 272K. Above 272K: input $4.00, cached $0.20, output $15.00. Cache write $2.50. Source, checked Source Oct 18, 2026
Claude Sonnet 5.5Anthropic (Claude)	$2	$10	$12
Details Provider Anthropic (Claude) Cached input / 1M tokens $0.10 Context 1M Note Same rate across the full 1M context. Cache write: 5 min $2.50, 1 hour $4. Source, checked Source Oct 18, 2026
Gemini 4 ArgonGoogle (Gemini API)	$2	$10	$12
Details Provider Google (Gemini API) Cached input / 1M tokens $0.10 Context 1.05M Note Price valid through 2026-11-30. From 2026-12-01: input $4.00, output $20.00, cached $0.20. Implicit caching only. Output includes thinking tokens. Source, checked Source Oct 18, 2026
Claude Opus 5.5Anthropic (Claude)	$4	$20	$24
Details Provider Anthropic (Claude) Cached input / 1M tokens $0.20 Context 1M Note Same rate across the full 1M context. Cache write: 5 min $5, 1 hour $8. Fast mode: input $8, output $40. Source, checked Source Oct 18, 2026
GPT-6 AstraOpenAI	$10	$50	$60
Details Provider OpenAI Cached input / 1M tokens $1 Context 1.05M Note Short-context rate (input up to 272K). Above 272K: input $20.00, cached $2.00, output $75.00. Cache write $12.50. Top model in the official lineup. Source, checked Source Oct 18, 2026
Claude Fable 5.1Anthropic (Claude)	$10	$50	$60
Details Provider Anthropic (Claude) Cached input / 1M tokens $0.25 Context 1M Note Same rate across the full 1M context. Cache write: 5 min $12.50, 1 hour $20. Source, checked Source Oct 18, 2026

Notes on the table

"1M in + 1M out" is the input price plus the output price: the cost of sending 1M tokens and getting 1M tokens back.
Reasoning ("thinking") tokens are billed as output on models that use them, so real output counts can be higher than the visible answer.
Long-context surcharges and the higher tiers of tiered pricing are not used.

Prices by provider

Pricing tables for other AI tools

Links marked "Affiliate link" pay us a commission if you subscribe. They do not change the order of the tables, which is sorted by price. Prices are as published by each vendor (most exclude sales tax and VAT). Check the vendor's page before you buy.

How this data is collected: Prices are read by hand from each vendor's official pricing page and stored with the date and source URL. Values we could not confirm there are left blank. Last checked Oct 18, 2026.