# Tokenando > Independent pricing and economics intelligence for AI models. Tokenando tracks per-token API prices for 653 AI models across 66 providers, with cost calculators, side-by-side comparisons, benchmark leaderboards, a dated price-change log, daily-settled market indices, and plain-English guides to how AI billing works. Every price is the rate a provider publishes, recorded with the date it was checked. Prices checked 3 September 2026 as of this file. Where a value is not published we leave it empty rather than estimate it. Three provenance rules matter if you are citing this data: - **Published prices** are input, output and cached-input rates, taken from provider pricing pages or their router listing. These are facts about what a provider charges. - **Blended rates and job costs are calculated by us**, not published by anyone. They weight input and output into one figure (70/30 by default) so models can be ranked on one axis. Always label them as derived. - **Benchmark scores are as reported** by the model's provider or a named public leaderboard, and are not independently re-run. Scores from different benchmark variants (SWE-bench Verified vs SWE-bench Pro, for example) are not comparable with each other. ## Pricing data - [AI model prices](https://tokenando.ai/pricing): live per-token input, output and blended rates for all 653 tracked models, sortable and filterable - [Model directory](https://tokenando.ai/models): every tracked model grouped by provider, with pricing and context window - [Provider directory](https://tokenando.ai/providers): all 66 providers with model counts, price ranges and flagship models - [Price change log](https://tokenando.ai/pricing/changes): every recorded change to a tracked rate, with the old value, the new value and the date, plus how many were cuts against increases - [Pricing JSON feed](https://tokenando.ai/api/prices): machine-readable current prices for every tracked model - [Pricing API documentation](https://tokenando.ai/developers): how to read the JSON feed, what each field means, and the attribution terms for citing these numbers - [Embeddable price table](https://tokenando.ai/widgets): a free live price widget for other sites, carrying its own attribution - [Methodology](https://tokenando.ai/about#methodology): how prices are sourced and verified, and how the blended rate is calculated ## Indices - [Tokenando Indices](https://tokenando.ai/indices): daily settled reference indices: TTPI (blended $/MTok per lab), TCPI ($/GPU-hr per GPU) and TIS (inference spread), all with open methodology - [TTPI — Token Price Index](https://tokenando.ai/indices/ttpi): volume-weighted blended USD per million tokens, per AI lab per day - [TCPI — Compute Price Index](https://tokenando.ai/indices/tcpi): weighted-median GPU rental price per GPU-hour from disclosed sources - [TIS — Inference Spread](https://tokenando.ai/indices/tis): lab API price minus verified self-hosted inference cost, plus the Frontier Premium ratio - [Index methodology](https://tokenando.ai/methodology): formulas, constants, source weights, exclusion rules, throughput citations and changelog — the settle computes from this same config - [Indices JSON API](https://tokenando.ai/api/v1/catalog): free keyless API: /api/v1/ttpi, /api/v1/tcpi, /api/v1/tis, /api/v1/frontier-premium ## Tools - [AI token calculator](https://tokenando.ai/calculator): count tokens in a prompt and estimate per-request and monthly cost for any model - [Model comparison tool](https://tokenando.ai/compare): compare up to four models side by side on price, context and benchmarks ## Rankings Computed from published prices and sourced benchmark scores. 16 lists, each stating its ranking method on the page. - [All rankings](https://tokenando.ai/best): directory of every ranked list - [Cheapest AI Models](https://tokenando.ai/best/cheapest-ai-models): The cheapest AI models ranked by blended cost per million tokens, from provider-published input and output rates. Updated daily - [Cheapest AI Models for RAG](https://tokenando.ai/best/cheapest-ai-models-for-rag): The cheapest AI models for RAG and long-context workloads, ranked on an input-heavy 80/20 weighting of published token prices - [Cheapest AI Models for Agents](https://tokenando.ai/best/cheapest-ai-models-for-agents): The cheapest AI models for agentic workloads, ranked on an output-heavy 20/80 weighting, where generation, not context, dominates cost - [Cheapest AI Models for Chat](https://tokenando.ai/best/cheapest-chat-models): The cheapest AI models for chat and assistant workloads, ranked on an even 50/50 weighting of published input and output token prices - [Cheapest Frontier AI Models](https://tokenando.ai/best/cheapest-frontier-models): The cheapest frontier AI models (the most capable tier from each major lab) ranked by blended cost per million tokens - [Cheapest Long-Context AI Models](https://tokenando.ai/best/cheapest-long-context-models): The cheapest AI models with a context window of 200K tokens or more, ranked by blended cost per million tokens - [Cheapest Vision AI Models](https://tokenando.ai/best/cheapest-vision-models): The cheapest AI models that accept image input, ranked by blended cost per million tokens from published provider rates - [Cheapest Embedding Models](https://tokenando.ai/best/cheapest-embedding-models): The cheapest embedding models for vector search and RAG, ranked by published input price per million tokens ## Benchmarks 6 leaderboards, each showing the token price beside every score. - [All benchmark leaderboards](https://tokenando.ai/benchmarks): directory, plus how to read benchmark scores safely - [SWE-bench Leaderboard](https://tokenando.ai/benchmarks/swe-bench): SWE-bench Verified leaderboard: AI models ranked by real GitHub issues resolved, with API pricing beside every score - [GPQA Leaderboard](https://tokenando.ai/benchmarks/gpqa): GPQA Diamond leaderboard: AI models ranked on graduate-level science reasoning, with token prices beside every score - [MMLU Leaderboard](https://tokenando.ai/benchmarks/mmlu): MMLU leaderboard: AI models ranked by multitask academic knowledge across 57 subjects, with API pricing alongside - [MATH Leaderboard](https://tokenando.ai/benchmarks/math): MATH benchmark leaderboard: AI models ranked on competition mathematics, with token prices beside every score - [HumanEval Leaderboard](https://tokenando.ai/benchmarks/humaneval): HumanEval leaderboard: AI models ranked by Python code generation accuracy, with API pricing beside every score - [LMArena Leaderboard](https://tokenando.ai/benchmarks/lmarena): LMArena Elo leaderboard: AI models ranked by blind head-to-head human preference, with token prices alongside ## Head-to-head comparisons 341 model matchups and 20 provider matchups, each with a full price table and a computed verdict. Examples: - [All matchups](https://tokenando.ai/vs): directory of every head-to-head comparison - [Claude Fable 5 vs DeepSeek-R1](https://tokenando.ai/vs/anthropic-claude-fable-5-vs-deepseek-deepseek-r1): price, context and benchmark comparison - [Claude Fable 5 vs DeepSeek-V3](https://tokenando.ai/vs/anthropic-claude-fable-5-vs-deepseek-deepseek-v3): price, context and benchmark comparison - [Claude Fable 5 vs DeepSeek-V4-Flash](https://tokenando.ai/vs/anthropic-claude-fable-5-vs-deepseek-deepseek-v4-flash): price, context and benchmark comparison - [Claude Fable 5 vs Gemini 2.5 Flash](https://tokenando.ai/vs/anthropic-claude-fable-5-vs-google-gemini-2-5-flash): price, context and benchmark comparison - [Claude Fable 5 vs Gemini 2.5 Pro](https://tokenando.ai/vs/anthropic-claude-fable-5-vs-google-gemini-2-5-pro): price, context and benchmark comparison - [Claude Fable 5 vs Gemini 3.1 Pro](https://tokenando.ai/vs/anthropic-claude-fable-5-vs-google-gemini-3-1-pro): price, context and benchmark comparison - [OpenAI vs Anthropic](https://tokenando.ai/vs/openai-vs-anthropic): vendor-level comparison of catalogue, price floor and ceiling ## Guides 17 explainers on AI cost mechanics, written for readers who pay the bill. - [All guides](https://tokenando.ai/learn): directory of every guide - [How AI API Pricing Works](https://tokenando.ai/learn/how-ai-api-pricing-works): AI APIs bill per token, charge more for output than input, and range over 1,900× in price. Here is what actually drives the number on your invoice - [Input vs Output Tokens: Why the Split Matters](https://tokenando.ai/learn/input-vs-output-tokens): Output tokens cost 2× more than input on the median AI model, and 3× or more on 121 of them. Why that asymmetry decides which model is cheapest for you - [Blended Cost Explained](https://tokenando.ai/learn/blended-cost-explained): A blended rate collapses input and output prices into one number for ranking. Useful for shortlisting, misleading for budgeting, and here is why - [Cost per 1,000 Tokens vs per Million](https://tokenando.ai/learn/cost-per-1000-tokens-vs-per-million): AI prices moved from per-1,000 to per-million tokens. The conversion is a factor of 1,000, and mixing the two units is a common budgeting error - [How to Calculate AI API Costs](https://tokenando.ai/learn/how-to-calculate-ai-api-costs): The arithmetic for estimating LLM API spend, a worked example, and the four mistakes that make most estimates too low - [How Context Windows Affect Cost](https://tokenando.ai/learn/how-context-windows-affect-cost): A large context window is a capacity, not a bill. What costs money is filling it, and several providers charge a higher rate past a threshold - [Prompt Caching: When It Saves 90%](https://tokenando.ai/learn/prompt-caching-cost-savings): Cached input typically costs 10% of the standard rate, but cache writes cost more than a normal request. The break-even maths and when caching pays - [Batch API Pricing Explained](https://tokenando.ai/learn/batch-api-pricing-explained): Batch processing halves both input and output rates in exchange for asynchronous delivery. No quality trade-off, which makes it unusually easy to justify - [Free AI APIs vs Paid: What You Give Up](https://tokenando.ai/learn/free-tier-vs-paid-api): Genuinely free AI models exist. The constraints are rate limits, availability and support rather than output quality, which changes where they fit - [Open-Weights vs API: The Real Cost Comparison](https://tokenando.ai/learn/open-source-vs-api-costs): Self-hosting an open-weights model trades a per-token bill for a GPU bill. Where the crossover sits depends almost entirely on utilisation - [Estimating RAG Costs](https://tokenando.ai/learn/estimating-rag-costs): Retrieval-augmented generation is input-heavy, so the input rate dominates. Embedding is the small line; resent context is the large one - [What Agentic Workloads Actually Cost](https://tokenando.ai/learn/agentic-workload-costs): Agents multiply output tokens, the expensive side of every price sheet, and reasoning models bill for thinking you never see - [How to Reduce LLM API Costs](https://tokenando.ai/learn/how-to-reduce-llm-api-costs): The levers that actually cut AI API spend, ranked by impact, starting with the one that dwarfs all the others - [Are AI Prices Actually Falling?](https://tokenando.ai/learn/ai-price-trends): We have logged every tracked price change since May. Cuts outnumber rises, but the average move is close to flat: prices churn rather than fall - [How to Choose an AI Model](https://tokenando.ai/learn/choosing-an-ai-model): A decision order that works: rule out on capacity, shortlist on price, break the tie on evidence, then verify on your own traffic - [Why AI Benchmark Scores Do not Compare](https://tokenando.ai/learn/why-ai-benchmark-scores-dont-compare): Two AI models can post benchmark scores that look comparable and are not. Here is what breaks comparability, and how to read scores safely - [AI Subscription vs API: Which Costs Less](https://tokenando.ai/learn/ai-subscription-vs-api-cost): A $20 monthly chat subscription buys roughly 25 million tokens at median API rates, or 1.6 million on a frontier model. Here is where the line falls ## Reference - [AI and LLM glossary](https://tokenando.ai/glossary): 63 terms defined in plain language, from tokens and context windows to MoE and RLHF - [State of the AI market](https://tokenando.ai/context): aggregate pricing data by category, tier and provider - [AI company map](https://tokenando.ai/companies): profiles of model providers, chipmakers, clouds and datacentre operators - [Who's Who in AI](https://tokenando.ai/people): profiles of the founders, executives and researchers shaping the field ## Optional - [AI news feed](https://tokenando.ai/news): curated industry news, refreshed continuously - [AI Funding & Capital News](https://tokenando.ai/news/topic/ai-funding): coverage volume, publisher spread and recent headlines from our archive - [AI Infrastructure News](https://tokenando.ai/news/topic/ai-infrastructure): coverage volume, publisher spread and recent headlines from our archive - [AI Model Release News](https://tokenando.ai/news/topic/ai-models): coverage volume, publisher spread and recent headlines from our archive - [AI Regulation & Policy News](https://tokenando.ai/news/topic/ai-regulation): coverage volume, publisher spread and recent headlines from our archive - [Open-Source AI News](https://tokenando.ai/news/topic/open-source-ai): coverage volume, publisher spread and recent headlines from our archive - [AI daily brief](https://tokenando.ai/brief): human-reviewed daily roundup - [The Compute](https://tokenando.ai/compute): weekly editorial briefing on AI market economics - [About Tokenando](https://tokenando.ai/about): who publishes this, and the independence policy