Skip to main content
Gradually LLM Price Index

What do LLM tokens really cost?

The Gradually LLM Price Index makes API prices comparable. You get a transparent market median from a curated basket of publicly priced text and code models, plus dedicated indexes for small, mid-range, and high-end models.

82 models and variants16 provider groups35 official sources410 endpoint tariffs7 separate price catalogs

Current market median

Methodology v2.2

$1.30

per 1M tokens

Small models · 22

$0.263

Mid-range models · 28

$1.40

High-end models · 32

$3.25

Price sources verified from July 30, 2026 to July 31, 2026. 75% input, 25% output.

Your scenario

Adjust the index to match your usage

Choose input prices, output prices, or a blended price. The entire index recalculates instantly.

Price view

Input and output prices are weighted by your token mix.

Share of input tokens

75%
75% Input25% Output

Token volume

Cost at the market median

$130.19

for the selected token volume. Excludes cache, batch, volume discounts, request fees, and search fees.

Price range

How far apart the models in the index really are

Each bar shows the blended list price per 1M tokens at the selected input share.

GPT-5.5 Pro

OpenAI · High-end models

$67.50

GPT-5.4 Pro

OpenAI · High-end models

$67.50

GPT-5 pro

OpenAI · High-end models

$41.25

o3-pro

OpenAI · High-end models

$35.00

o1

OpenAI · High-end models

$26.25

Claude Fable 5

Anthropic · High-end models

$20.00

Claude Mythos 5

Anthropic · High-end models

$20.00

ChatGPT chat-latest

OpenAI · Mid-range models

$11.25

GPT-5.6 Sol

OpenAI · High-end models

$11.25

GPT-5.5

OpenAI · High-end models

$11.25

Claude Opus 5

Anthropic · High-end models

$10.00

Claude Opus 4.7

Anthropic · High-end models

$10.00

Claude Opus 4.6

Anthropic · High-end models

$10.00

Claude Opus 4.5

Anthropic · High-end models

$10.00

Claude Sonnet 4.6

Anthropic · Mid-range models

$6.00

Claude Sonnet 4.5

Anthropic · Mid-range models

$6.00

Sonar Pro

Perplexity · Mid-range models

$6.00

Kimi K3

Moonshot AI · High-end models

$6.00

Small modelsMid-range modelsHigh-end modelsMedian $1.30

What the number means

A market benchmark you can recalculate

The index is deliberately not a mysterious financial ticker. You can inspect every model, price, and calculation rule. Change the token mix and the benchmark moves for a reason you can see.

Lowest-cost model

$0.061

Nova Micro 1.0

Highest-cost model

$67.50

GPT-5.5 Pro

Price multiple

1,102×

between the lowest-cost and highest-cost model

View every model, price, and source

OpenRouter pricing layer

Every OpenRouter provider in view

In addition to direct developer prices, the index tracks the complete OpenRouter provider catalog. Every available endpoint tariff, provider, and price range is evaluated for each mapped basket model.

OpenRouter data verified on July 31, 2026. The complete OpenRouter catalog contains 364 models.

101

providers in the catalog

52

providers serving basket models

410

endpoint tariffs

78/82

basket models mapped

OpenRouter tariffs expand source coverage but do not feed into the main index. Multiple providers, regions, and service tiers can offer the same model. Counting them as separate models would distort the market median. Price ranges use the default prices reported by OpenRouter and exclude conditional overrides for long contexts or time windows.

View every OpenRouter provider and price source

Market-wide source layer

7 price catalogs, 176 provider sources

OpenRouter covers only part of the market. Gradually therefore cross-checks the index against models.dev, Pydantic GenAI Prices, Portkey Models, ModelMeta, Tokrates, the LiteLLM Model Cost Map, and the Helicone Model Registry. The directory below exposes every provider source in models.dev, including model counts, complete token price coverage, and original documentation.

Catalog snapshots retrieved on July 31, 2026.

7

separate price catalogs

176

providers on models.dev

433

provider entries in total

13,036

complete price entries

The external catalogs form a validation and research layer. They contain many overlaps, variants, and resold model endpoints. Their entries are therefore not added together or counted as extra models in the main median. Directly verified developer prices remain the basis of the index.

Catalog entries, not unique models

models.dev

Providers
176
Model entries
5,908
Fully priced
5,504
Open catalog

Pydantic GenAI Prices

Providers
34
Model entries
1,285
Fully priced
1,218
Open catalog

Portkey Models

Providers
45
Model entries
3,026
Fully priced
2,909
Open catalog

ModelMeta

Providers
26
Model entries
820
Fully priced
657
Open catalog

Tokrates

Providers
8
Model entries
139
Fully priced
139
Open catalog

LiteLLM Model Cost Map

Providers
123
Model entries
2,985
Fully priced
2,498
Open catalog

Helicone Model Registry

Providers
21
Model entries
111
Fully priced
111
Open catalog
View all 176 models.dev provider sources

Methodology

How the Gradually LLM Price Index is calculated

The calculation is intentionally simple. You can verify the benchmark yourself and see immediately what it captures and what it leaves out.

01

Define the model basket

The index includes publicly priced text and code models that are available directly through their developers' APIs and marked as current in Gradually's central pricing database. Specialized image, audio, embedding, and moderation models are excluded.

02

Assign product classes

Each model is classified as small, mid-range, or high-end based on its role within the provider's product family. Undisclosed or estimated parameter counts are not used.

03

Balance providers fairly

A separate median is calculated for each provider first. The overall index and three class indexes are then calculated from those provider medians, so a provider with many variants does not carry more weight.

The formula

Model price = input price × input share + output price × output share. Provider value = median model price for one provider. Index = median of all provider values.

What the index does not capture

The index uses the public base tier of API list prices. It excludes context-based pricing tiers, regional price differences, negotiated discounts, caching, batch processing, request fees, search queries, citation and reasoning-token surcharges, self-hosted models, taxes, and real provider usage shares. The curated basket covers major providers, but not every API and hosting service in the market. The index is a comparison benchmark, not a forecast or financial index.

Want to calculate your actual API costs?

The index shows the market. For one specific model and your exact workflow, the full API cost calculator is the better tool.

Frequently asked questions about the LLM Price Index

Short, honest answers