Realm by Rook
AI product engineeringProduct designAI implementationWorkflow automationSystems integrationDomain strategyAI product engineeringProduct designAI implementationWorkflow automationSystems integrationDomain strategy
Developer & Technical Tools

LLM Price Comparison

Live pricing table for every frontier LLM — GPT, Claude, Gemini, Llama, DeepSeek, Mistral

Prices last verified April 2026. Official vendor pricing pages take precedence if they diverge.
Cheapest (input + output)
GPT-5 nano
$0.05 in · $0.4 out
Largest context window
Llama 4 Maverick
10,000,000 tokens
Model Vendor Input $ / 1M Output $ / 1M Context Released Scenario $
GPT-5 nano
text
OpenAI$0.05$0.40272,0002025-08-07$0.150
Gemini 2.5 Flash-Lite
text · vision
Google$0.10$0.401,000,0002025-07-22$0.200
Llama 4 Scout
text · vision
Meta$0.11$0.3410,000,0002025-04-05$0.195
GPT-5 mini
text · vision
OpenAI$0.25$2.00272,0002025-08-07$0.750
Llama 4 Maverick
text · vision
Meta$0.27$0.8510,000,0002025-04-05$0.483
DeepSeek V3.1
text
DeepSeek$0.27$1.10128,0002025-08-21$0.545
Gemini 2.5 Flash
text · vision · audio · video
Google$0.30$2.501,048,5762025-06-17$0.925
Mistral Medium 3
text
Mistral$0.40$2.00128,0002025-05-07$0.900
DeepSeek R1 (reasoning)
text
DeepSeek$0.55$2.1964,0002025-01-20$1.098
Claude Haiku 4.5
text · vision
Anthropic$1.00$5.00200,0002025-10-01$2.250
GPT-5
text · vision
OpenAI$1.25$10.00272,0002025-08-07$3.750
Gemini 2.5 Pro
text · vision · audio · video
Google$1.25$10.002,097,1522025-06-17$3.750
GPT-4.1
text · vision
OpenAI$2.00$8.001,047,5762025-04-14$4.000
o3 (reasoning)
text · vision
OpenAI$2.00$8.00200,0002025-04-16$4.000
Mistral Large 2
text
Mistral$2.00$6.00128,0002024-11-18$3.500
Claude Sonnet 4.6
text · vision
Anthropic$3.00$15.001,000,0002026-01-22$6.750
Grok 4
text · vision
xAI$3.00$15.00256,0002025-07-09$6.750
Claude Opus 4.7
text · vision
Anthropic$15.00$75.001,000,0002026-03-10$33.750

How to use

  1. Sort the table by column
  2. Filter by vendor
  3. Enter your scenario tokens for per-model cost

Use cases

  • Model selection
  • Cost forecasting
  • Vendor comparison

About the LLM Price Comparison

Choosing an LLM vendor has become a 7-figure decision for many companies, yet prices, context windows, and capabilities change quarterly. This comparison table gives you a single-pane view across every frontier model, with a scenario calculator so you can instantly see what your workload would cost on each. Prices are verified monthly against vendor pricing pages.

Features

  • 17+ models from OpenAI, Anthropic, Google, Meta, DeepSeek, Mistral, xAI
  • Sortable by price, context window, release date
  • Vendor filter
  • Scenario calculator: plug in input/output token counts, see each model's cost
  • Updated monthly — current as of April 2026

How it works

  1. Review the table (sort by the column that matters most to you).
  2. Filter by vendor if you're evaluating a specific provider.
  3. Enter your scenario's input + output token counts to see per-call cost on every model.

Frequently asked questions

How often is this updated?

+

Monthly. We verify against each vendor's official pricing page at the start of each month.

Does this include Enterprise / committed-use discounts?

+

No — all prices shown are standard pay-as-you-go API tier. Volume discounts (OpenAI >$100k/mo, Anthropic custom contracts, AWS Bedrock committed use) can cut 20-40% off these rates.

What about cached-input pricing?

+

Cached-input tokens (repeated system prompts or context) price at ~10-25% of standard input. Many workloads save 50%+ with caching — a big lever we recommend profiling.

Are the open-weight models (Llama, DeepSeek) free?

+

The weights are free to run yourself. The prices we show are for managed endpoints (Together, Fireworks, Groq, etc. — representative of market).

Do these prices include vision / audio tokens?

+

No — images and audio have separate per-token or per-request pricing that varies by model. Text-only here.