optimnow-io Llm Models API

openapi: 3.2.0
info:
  title: OptimToken Public Llm Models API
  version: '2.7'
  x-schema-versions:
    /api/llm-models: '2.7'
    /api/pricing: '1.3'
  description: 'Free, no-auth JSON API behind optimtoken.optimnow.io. Compare LLM API pricing (incl. batch and prompt-cache rates, Arena ELO quality scores) and cloud compute instance pricing across 7 providers.


    Field provenance is marked in each description: [published] = taken directly from the provider/source API; [derived] = computed from published values with a documented formula; [estimated] = curated fallback value, not a quote.


    info.version is the schemaVersion of /api/llm-models; /api/pricing is versioned separately (x-schema-versions), and each response states its own in meta.schemaVersion.


    Licence: the data is published under Creative Commons Attribution 4.0 (CC BY 4.0). When you use it, cite https://optimtoken.optimnow.io/ as the source. Data is provided as-is for comparison purposes; verify final numbers with official provider pricing pages.'
  contact:
    name: OptimNow
    url: https://www.optimnow.io/contact
  license:
    name: CC BY 4.0
    url: https://creativecommons.org/licenses/by/4.0/
servers:
- url: https://optimtoken.optimnow.io
tags:
- name: Llm Models
paths:
  /api/llm-models:
    get:
      operationId: listLlmModels
      summary: LLM pricing catalogue
      description: All tracked LLM models with USD prices per 1 million tokens. Refreshed at most every 24h (CDN cache). Sorted by Arena ELO descending, then release date.
      parameters:
      - name: provider
        in: query
        schema:
          type: string
        description: Exact provider name, case-insensitive (e.g. Anthropic, OpenAI, Google).
      - name: model
        in: query
        schema:
          type: string
        description: Case-insensitive substring match on the model name (e.g. 'sonnet').
      - name: limit
        in: query
        schema:
          type: integer
          minimum: 1
        description: Maximum number of models returned, applied after filters.
      responses:
        '200':
          description: Model list
          content:
            application/json:
              schema:
                type: object
                properties:
                  models:
                    type: array
                    items:
                      $ref: '#/components/schemas/LLMModel'
                  meta:
                    $ref: '#/components/schemas/LLMMeta'
      tags:
      - Llm Models
components:
  schemas:
    LLMMeta:
      type: object
      description: Referenced by this document but never defined in it. API Evangelist added this empty placeholder so the document resolves; the shape is unknown and is NOT a claim about the API.
      x-ae-placeholder: true
    LLMModel:
      type: object
      properties:
        provider:
          type: string
          description: '[published] Vendor display name (e.g. Anthropic, OpenAI).'
        model:
          type: string
          description: '[published] Model display name (e.g. Claude Sonnet 5).'
        parameters:
          type: string
          description: '[published/estimated] Parameter count where disclosed (e.g. 70B, 1T); curated for major models, extracted from names/descriptions otherwise.'
        inputPricePer1M:
          type: number
          description: '[published] USD per 1M input tokens, list price, 4-decimal precision.'
        outputPricePer1M:
          type: number
          description: '[published] USD per 1M output tokens, list price.'
        batchInputPricePer1M:
          type: number
          description: '[published] Batch API input price where the provider publishes one (typically -50%). Absent = not published.'
        batchOutputPricePer1M:
          type: number
          description: '[published] Batch API output price. Absent = not published.'
        cachedInputPricePer1M:
          type: number
          description: '[published] Prompt-cache READ price per 1M tokens (typically ~10% of input). Absent = not published.'
        cacheWritePricePer1M:
          type: number
          description: '[published] Prompt-cache WRITE price per 1M tokens at the default (roughly 5-minute) cache lifetime, schemaVersion 2.3+. Typically 1.25x input: writing a prefix into the cache costs MORE than sending it uncached, so caching only pays once the prefix is re-read. The dearer 1-hour write some vendors sell is not carried. Some vendors publish a figure BELOW the input rate here (Google, 2026-09-21); this site only ever adds the premium above input, so such a value changes no cost figure. Absent = not published, which does NOT mean free: assume a write costs the ordinary input rate. Dropped for models whose price is corrected by hand.'
        contextWindow:
          type: string
          description: '[published] Context window (e.g. 200K, 1M).'
        outlierListing:
          type: boolean
          description: '[derived] schemaVersion 2.6+. True when the price this row publishes is an OUTLIER LISTING: one host''s rate with input near free (at most 35% of the median host input) and an output/input ratio at least 4x the median, the shape a few small resellers list.'