LLM API Pricing

Up-to-date pricing for 300+ LLM APIs in a single JSON file.

Powered by BenchGecko -- The Data Layer of the AI Economy.

---

Why This Exists

Every LLM provider has different pricing pages, different formats, different units. This repo gives you one canonical JSON file with standardized pricing for every major LLM API, updated weekly from the OpenRouter aggregator.

Use it to build cost calculators, comparison tools, billing dashboards, or just to answer "how much does GPT-4o cost again?"

For full benchmark scores, performance comparisons, and provider analytics, see benchgecko.ai.

---

Quick Start

Fetch the latest pricing

# Raw JSON (346 models, ~200KB)
curl -sL https://raw.githubusercontent.com/BenchGecko/llm-pricing/main/pricing.json | python3 -m json.tool | head -50

# Grouped by provider
curl -sL https://raw.githubusercontent.com/BenchGecko/llm-pricing/main/pricing-by-provider.json

JavaScript / TypeScript

const res = await fetch(
  'https://raw.githubusercontent.com/BenchGecko/llm-pricing/main/pricing.json'
);
const { models } = await res.json();

// Find cheapest model with 100K+ context
const cheap = models
  .filter(m => !m.is_free && m.context_window >= 100000)
  .sort((a, b) => a.input_per_million - b.input_per_million);

console.log(cheap[0]);
// { id: "...", name: "...", input_per_million: 0.08, ... }

Python

import requests

data = requests.get(
    "https://raw.githubusercontent.com/BenchGecko/llm-pricing/main/pricing.json"
).json()

# Calculate cost for 1M input + 100K output tokens
for m in data["models"]:
    if "gpt-4" in m["id"]:
        cost = m["input_per_million"] + (m["output_per_million"] * 0.1)
        print(f"{m['name']}: ${cost:.2f}")

Install via npm

npm install @benchgecko/llm-pricing

const pricing = require('@benchgecko/llm-pricing');
console.log(pricing.models.length); // 346
console.log(pricing.models.find(m => m.id === 'openai/gpt-4o'));

Note: For always-fresh data, use the raw GitHub URL. The npm package is updated weekly.

---

Top Models by Price

Model	Provider	Input ($/1M tokens)	Output ($/1M tokens)	Context
Claude Opus 4.6	Anthropic	$5.00	$25.00	1000K
Claude Sonnet 4.6	Anthropic	$3.00	$15.00	1000K
Claude Sonnet 4.5	Anthropic	$3.00	$15.00	1000K
Claude Haiku 4.5	Anthropic	$1.00	$5.00	200K
DeepSeek V3 0324	DeepSeek	$0.20	$0.77	164K
R1	DeepSeek	$0.70	$2.50	64K
Gemini 2.5 Pro	Google DeepMind	$1.25	$10.00	1049K
Gemini 2.5 Flash	Google DeepMind	$0.30	$2.50	1049K
Gemini 2.0 Flash	Google DeepMind	$0.10	$0.40	1049K
Llama 4 Maverick	Meta	$0.15	$0.60	1049K
Llama 4 Scout	Meta	$0.08	$0.30	328K
Mistral Large 2411	Mistral AI	$2.00	$6.00	131K
GPT-4.1	OpenAI	$2.00	$8.00	1048K
GPT-4.1 Mini	OpenAI	$0.40	$1.60	1048K
GPT-4.1 Nano	OpenAI	$0.10	$0.40	1048K
GPT-4o	OpenAI	$2.50	$10.00	128K
GPT-4o-mini	OpenAI	$0.15	$0.60	128K
o3	OpenAI	$2.00	$8.00	200K
o3 Mini	OpenAI	$1.10	$4.40	200K
o4 Mini	OpenAI	$1.10	$4.40	200K

Full pricing for all 346 models available in pricing.json.

Want to compare performance too? See how these models score on 40+ benchmarks at benchgecko.ai/compare

---

Data Schema

Each model in pricing.json has this structure:

{
  "id": "openai/gpt-4o",
  "name": "GPT-4o",
  "provider": "OpenAI",
  "input_per_million": 2.5,
  "output_per_million": 10.0,
  "context_window": 128000,
  "max_output": 16384,
  "is_free": false,
  "modalities": ["file", "image", "text"],
  "type": "chat"
}

Field	Type	Description
id	string	OpenRouter model identifier (e.g., openai/gpt-4o)
name	string	Human-readable model name
provider	string	Company/org that created the model
input_per_million	number	Cost in USD per 1 million input tokens
output_per_million	number	Cost in USD per 1 million output tokens
context_window	number | null	Maximum context length in tokens
max_output	number | null	Maximum output/completion tokens
is_free	boolean	Whether the model is free to use
modalities	string[]	Supported modalities (text, image, audio, video, file)
type	string	Model type: chat, image, embedding, speech

Files

File	Description
pricing.json	Flat list of all models, sorted by provider then name
pricing-by-provider.json	Same data grouped by provider name

---

Related Resources

• BenchGecko — Full model rankings, benchmark scores, and provider analytics
• BenchGecko Compare — Side-by-side model comparison with pricing + performance
• BenchGecko API — Free REST API for model data, benchmarks, and pricing
• Awesome LLM Benchmarks — Curated list of 60+ AI benchmarks

---

Data Source

Pricing data is sourced from the OpenRouter API and updated weekly. OpenRouter aggregates pricing from all major LLM providers into a single API.

Prices reflect the OpenRouter pass-through rates, which match or closely track the official provider pricing.

For full benchmark comparisons, performance scores, and provider analytics, visit benchgecko.ai.

---

License

MIT -- use this data however you want. Attribution appreciated but not required.

If you build something cool with this data, let us know at benchgecko.ai or @BenchGecko on Twitter.