# OptimToken by OptimNow

> Compare LLM pricing and cloud compute pricing with business metrics, technical filters, and live provider data.

Primary site: https://optimtoken.optimnow.io/
Compute page: https://optimtoken.optimnow.io/compute
Per-model pages: https://optimtoken.optimnow.io/models/{provider}/{model}, one
static HTML page per model (full pricing incl. batch/cache, use-case costs),
readable without JavaScript; complete list in https://optimtoken.optimnow.io/sitemap.xml
Models by maker: https://optimtoken.optimnow.io/models, every model in the
catalogue as a plain link, grouped by maker. A maker with 3 models or
more has a page of its own at /models/{maker}: 24 of the 44 makers
on 2026-10-28
Price barometer: https://optimtoken.optimnow.io/barometer, every change of 1% or
more in the price we quote for an LLM over the last 7 days, with the blended
30/70 price before and after. Two parts: price changes, and moves flagged as a
change of quoted host or an outlier listing (most models are sold by several
hosts at different rates, and the site quotes one of them). One permanent page
per week at /barometer/{year}-w{week}, listed at
https://optimtoken.optimnow.io/barometer/weeks, and an RSS feed at
https://optimtoken.optimnow.io/barometer/feed.xml
Guides index: https://optimtoken.optimnow.io/guides, all FinOps guides on LLM and
cloud pricing
Compare tool: https://optimtoken.optimnow.io/compare, any 2-4 models side by
side, plus prebuilt head-to-head pages at /compare/{model-a}-vs-{model-b}
Guides:
- [ARM against x86: what Graviton, Ampere and Axion save](https://optimtoken.optimnow.io/guides/arm-against-x86): The ARM discount clouds advertise is a discount against Intel. This guide measures it at matched vCPU and memory, against Intel and against AMD. Figures as JSON: https://optimtoken.optimnow.io/guides/arm-against-x86/figures.json
- [Batch API economics: when the -50% applies](https://optimtoken.optimnow.io/guides/batch-api-economics): Batch APIs cut LLM token prices roughly in half, but only for async workloads and only where a batch rate is published. Worked examples, current prices. Figures as JSON: https://optimtoken.optimnow.io/guides/batch-api-economics/figures.json
- [Vision, audio, agents: what each capability adds to the bill](https://optimtoken.optimnow.io/guides/capability-premium): What image input, audio and tool use add to an LLM bill, measured within each model maker's line-up and across the catalogue, and the shortlist each leaves.
- [Context windows: what a million tokens costs to fill](https://optimtoken.optimnow.io/guides/context-window-economics): A million-token context window is a spec-sheet capability and an invoice line. What one full prompt costs across the catalogue, and what re-reading it costs.
- [The cost of reasoning: what thinking models charge for output](https://optimtoken.optimnow.io/guides/cost-of-reasoning): Reasoning tokens are billed as output tokens, where prompt caching cannot reach them. What a thinking budget costs across the catalogue, at current prices.
- [How we compute cost per request](https://optimtoken.optimnow.io/guides/cost-per-request-methodology): The exact formulas behind the list-price and after-discounts columns: token profiles, cache and batch rules, what the numbers exclude and where they mislead.
- [How the models are scored: Arena quality, Efficiency, the FinOps badge and quality bars](https://optimtoken.optimnow.io/guides/how-models-are-scored): Where the quality score comes from, how the Efficiency score and the FinOps badge are calculated, and how a quality bar is set for a use case.
- [LLM API pricing statistics: cheapest and median](https://optimtoken.optimnow.io/guides/llm-pricing-statistics): How many LLM APIs carry a token price, which are the cheapest and the dearest, and the median price per million tokens by tier and by licence. Rebuilt daily.
- [Open weights against proprietary APIs: the cost gap in numbers](https://optimtoken.optimnow.io/guides/open-weights-vs-proprietary): Open-licensed models undercut proprietary APIs at matched quality across most of the range, but the open catalogue stops below the top tier. Live prices.
- [Prompt caching: the discount nobody budgets for](https://optimtoken.optimnow.io/guides/prompt-caching-economics): Cache reads bill at a fraction of the input price, but the saving that reaches the bill is far smaller than the headline. Break-even by use case, live prices.
- [A quality bar for each use case, and the cheapest model that clears it](https://optimtoken.optimnow.io/guides/quality-bar-price-floor): Decide per use case how good the answer has to be, set the bar on Arena, then buy the cheapest model that clears it. A method, with current prices.
- [Quality per dollar: the models that beat their price class](https://optimtoken.optimnow.io/guides/quality-per-dollar): A model's price class comes from its rate card, not its quality. Measured against Arena scores, most of the catalogue has a cheaper equal. Current prices.
- [Savings plans against reserved instances: what the flexibility costs](https://optimtoken.optimnow.io/guides/savings-plans-vs-reserved-instances): AWS and Azure each sell a flexible commitment and a locked one. What each takes off on-demand, the gap between the two and the utilisation where each pays off. Figures as JSON: https://optimtoken.optimnow.io/guides/savings-plans-vs-reserved-instances/figures.json
- [Bedrock, Vertex, Azure or direct: what the channel changes](https://optimtoken.optimnow.io/guides/where-to-run-a-model): The same model is sold by its model maker and by the hyperscalers. The token rate is identical on most comparisons; regions and commitment terms are not.
Sitemap: https://optimtoken.optimnow.io/sitemap.xml
Detailed reference: https://optimtoken.optimnow.io/llms-full.txt

## Summary

OptimToken helps teams compare large language model pricing and cloud infrastructure pricing in one web app. The root page focuses on LLMs. The compute page focuses on cloud instances.

## LLM pricing

- Compare hundreds of models when live OpenRouter data is available (exact count in the API's meta.total), with static fallback coverage in local development or API outages
- Model makers in the catalogue on 2026-10-28: 44 in all. The 12 with the most models are OpenAI, Alibaba, Google, Mistral, Meta, Zhipu, Anthropic, DeepSeek, MiniMax, Moonshot, ByteDance, xAI, and 32 others
- Business view shows efficiency score, unit cost per request, monthly budget estimates, and a FinOps Friendly badge
- Technical view shows input and output price per 1M tokens, context window, parameters, release date, and capabilities
- Price tier is set by list price alone and shown as Premium, Mid-range, Budget or Image (the API's `category` values Frontier, Mid-tier, Budget, Image). It is not a quality grade
- "Frontier" on the site means quality, not price: a model whose Arena score is within 25 points of the best ranked model: the 20 points under which Arena does not separate two models, plus 5. Shown as a chip and a filter in both views. A model with no Arena score is never called Frontier
- Use case presets: Support Ticket, Knowledge Q&A, Meeting Summary, Marketing Content, Coding Task, Invoice Processing, Call Summary, Agent Workflow
- FinOps Friendly requires all 4, on relative percentiles that recalibrate as the market moves:
  Arena ELO in the top 40% of ranked models, efficiency score in the top 30%, list price among
  the cheapest 70% of ranked models and not an outlier listing, and a stable release (no preview/beta/experimental label,
  not a Premium-tier model (API category "Frontier") released in the last 3 months,
  and no retirement announced by the model maker).
  The price gate reads list prices only: batch and cache discounts are conditional on the
  customer implementing them.

## Cloud compute pricing

- Compare AWS, Azure, GCP, DigitalOcean, OCI, OVH, and Alibaba Cloud
- Pricing tiers include on-demand, spot/preemptible, savings plans or CUDs, and reserved pricing where the provider supports them
- Filters cover provider, category, use case, processor family, operating system, region, vCPUs, and memory

## Machine-readable API (free, no auth, CORS enabled)

AI agents and applications are welcome to read the same JSON data the site uses.
Licence: the data is published under Creative Commons Attribution 4.0 (CC BY 4.0,
https://creativecommons.org/licenses/by/4.0/): cite https://optimtoken.optimnow.io/
as the source. Data is provided as-is for comparison; verify final numbers with
official provider pricing pages.

OpenAPI 3.1 spec (full schemas + field provenance): https://optimtoken.optimnow.io/openapi.json
MCP server (Model Context Protocol, streamable HTTP): https://optimtoken-mcp.optimnow.io/mcp
  Docs & connection guide: https://optimtoken.optimnow.io/mcp
  Tools: compare-llm-models, recommend-llm-model, compare-models-side-by-side,
  estimate-llm-cost, compare-compute-pricing.
  Registry name: io.github.optimnow/ai-pricing-hub

- LLM pricing JSON: https://optimtoken.optimnow.io/api/llm-models
  Query params: ?provider= (exact, case-insensitive), ?model= (substring), ?limit=
  Example: /api/llm-models?provider=anthropic&limit=10
  Returns { models: [...], meta }. Each model has: provider, model, inputPricePer1M,
  outputPricePer1M, batchInputPricePer1M, batchOutputPricePer1M (batch API, typically -50%),
  cachedInputPricePer1M (prompt-cache read price), cacheWritePricePer1M (prompt-cache
  WRITE price, default ~5-minute lifetime; absent = not published, not free),
  contextWindow, category, capabilities (the "Reasoning" tag means the model exposes a
  reasoning/thinking-budget parameter upstream, not that it is a dedicated reasoning
  model or a better one; most current models carry it. There is no "Code" tag:
  it was a price threshold and was removed in schemaVersion 2.5),
  releaseDate, eloScore (Arena), license. meta.timestamp is the snapshot time and
  meta.eloAsOf the date of the last Arena ELO refresh.
  meta.unpriced lists models OpenRouter carries with NO published token price (its
  anonymous "stealth" evaluation endpoints). They are not in models[] and are not
  free: the price is unpublished, prompts are retained by the anonymous
  provider, and every such model so far was withdrawn or renamed within weeks.
  Never quote a price for one, and never call it the cheapest option.
- Compute pricing JSON: https://optimtoken.optimnow.io/api/pricing
  Query params: ?region= (us-east | us-west | europe | asia-pacific),
  ?provider=aws,gcp (comma list), ?instanceType= (substring), ?limit=
  Example: /api/pricing?region=europe&provider=aws,azure&instanceType=large&limit=50
  Returns { instances: [...], meta }. Each instance has: provider, instanceType, os, vCPUs,
  memory, onDemandHourly, onDemandMonthly, spot, savingsPlan1yr/3yr, reserved1yr/3yr
  (reserved = EFFECTIVE HOURLY, term totals divided by term hours).
  meta.sources marks each provider live or static; meta.sourceRegions tells the actual
  priced region per provider: some public endpoints return a global list price.

## Data sources and freshness

- LLM pricing: OpenRouter API, refreshed at most every 24 hours (CDN cache), with a
  daily-regenerated static fallback
- LLM quality: Arena (formerly Chatbot Arena / LMArena) ELO scores, rebased WEEKLY from
  the Arena leaderboard by a GitHub Action (Mondays). Every mapped score is rewritten in one
  pass or none is: the site ranks on percentiles, so a partial refresh reads as market
  movement. meta.eloAsOf carries the leaderboard's own stamp, not the run date
- Cloud pricing: live provider APIs (AWS, Azure, GCP, DigitalOcean, OCI, OVH), refreshed
  at most every 24 hours; Alibaba Cloud uses curated static data