AI API Token Pricing Comparison

Compare pay-as-you-go API prices in USD per 1 million tokens, updated ; for monthly ChatGPT, Claude, Gemini, and Copilot plans, use subscription pricing, and verify rates directly before purchase because providers can change them without notice.

Showing 104 current models. .
Relative priceLowestLowerHigherHighestAutomatic log scale across current offers; input, cache and output are graded separately. Color shows cost, not quality.
Showing 104 grouped models from 134 offers · USD per 1M tokens
Last synced:

Sponsored links may earn us a commission at no extra cost to you. Affiliate status never changes model ordering.

All product names, logos, and brands are property of their respective owners and are used for identification purposes only.

AI model price scale

Every current canonical model at its cheapest active host. USD per 1M tokens.

Swipe horizontally to exploreTap a dot for prices
$0.05$0.1$0.5$1$5$10$50$100OpenAIAnthropicGooglexAIPerplexityMistralDeepSeekAlibabaZ.aiMoonshotMiniMaxMetaCohereNVIDIADeep CogitoEssential AIUSD per 1M tokens · log scale
Current modelFlagship tier104 models · 16 creators

All product names, logos, and brands are property of their respective owners and are used for identification purposes only.

aipricing.guru · updated 2026-08-25

Which AI API is cheapest right now? We track 215 models across 17 providers. The cheapest flagship is GPT-5.6 Luna at $0.20 per 1M input tokens; the absolute cheapest production model is Llama 3.1 8B Instruct at $0.02 per 1M input. The most expensive we track is GPT-5.4 Pro at $30.00 input / 180.00 output. Download the raw data as JSON.

Novita open-model API

One API for Llama, Qwen, DeepSeek and GLM — new accounts get $100 in sandbox credits for 90 days.

Get $100 sandbox credits

Affiliate disclosure: sponsored links may earn us a commission at no extra cost to you. Affiliate status never changes model ordering.

Qwen model family

Compare every current Qwen route

Separate Alibaba first-party aliases from managed Qwen hosts, with cache and long-context tiers.

Need exact math?

Use the token cost calculator

Enter your input/output token volume and estimate monthly spend before choosing a model.

Labs benchmark

Compare cost per correct answer

See which models turn token spend into correct reasoning, extraction, coding, and pricing answers.

Model pages

Open detailed model pricing

See per-model calculators, cached-token math, price history, and alternatives for high-demand models.

Subscription or API?

Compare monthly plans vs API usage

If you use AI through a chat app, calculate whether a subscription is cheaper than raw API tokens.

Frequently asked questions

Quick answers about API token pricing, freshness, and how to compare providers.

What does this AI API pricing comparison cover?

This page tracks pay-as-you-go API token pricing for 215 AI models across 17 providers — OpenAI, Anthropic, Google, DeepSeek, Mistral, xAI, Cohere, Groq, Together AI, Perplexity, and Meta Llama hosts. Each row shows input price, output price, cached-input price (where the provider offers prompt caching), and batch-discounted rates per million tokens. The data snapshot shown is from 2026-08-25.

How often is the pricing data updated?

Prices are verified and reconciled daily by an automated pipeline that pulls from each provider's official pricing page and cross-checks against at least two independent sources before publishing. New models and price changes typically appear here within hours of the provider's announcement. If a provider hasn't moved its prices, the snapshot date stays the same — the current dataset is from 2026-08-25.

Which AI API is the cheapest right now?

As of 2026-08-25, the absolute cheapest production model we track is Llama 3.1 8B Instruct at $0.02 per million input tokens. The cheapest flagship-class model — meaning a top-tier model from a major lab, not a small or distilled variant — is GPT-5.6 Luna at $0.20 per million input tokens. The most expensive model we track is GPT-5.4 Pro at $30.00 input / 180.00 output per 1M tokens.

How do I compare two specific AI providers like OpenAI and Anthropic?

Use the table search and filter controls above to narrow to a single provider, then sort by input or output price. For deeper provider-level pages with FAQs, methodology, and historical pricing, see /openai-pricing/, /anthropic-pricing/, /google-pricing/, /deepseek-pricing/, /mistral-ai-pricing/, /cohere-pricing/, or any other provider listed. For workload-level math, the token cost calculator at /calculators/token-cost/ runs your input/output volume against every model side-by-side.

Should I pay per token via API or subscribe to ChatGPT Plus or Claude Pro?

It depends on volume and access pattern. Monthly subscriptions like ChatGPT Plus, Claude Pro, and Gemini Advanced at roughly $20/month are usually cheaper than API access for chat-style usage under about 5 million input tokens per month. Production workloads, agents, and anything that needs API access almost always cost less pay-as-you-go on the API. The /subscription-vs-api/ tool computes the exact break-even point for your specific volume.

Can I download the AI API pricing data as JSON?

Yes. The full live dataset is published at /api/pricing.json and updated whenever this page is. The schema is { id, name, family, provider, pricing: { inputPerM, cachedInputPerM, outputPerM }, status } for all 215 tracked models. The AI Pricing Guru dataset is provided for informational use. You may use it for personal, editorial, research, educational, and internal business purposes with appropriate attribution to AI Pricing Guru. Commercial redistribution, resale, republishing at scale, inclusion in a competing public dataset/API, or use as the primary data source for a commercial pricing-comparison product requires prior written permission.

Token cost calculator → · Subscription vs API → · Download JSON → · Cost benchmarks → · Labs leaderboard →