LLM Cost Scanner

Compare the estimated monthly base-token cost of OpenRouter models from two aggregate token counts. The Actor reads the current public model catalog, ranks the least-expensive options, and returns one structured dataset item with source provenance and explicit limitations.

What you get

• A ranked list of up to 100 token-priced models.
• Input and output price per million tokens.
• Estimated monthly cost for your aggregate workload.
• Live-catalog timestamp, usable-model count, skipped-record count, and fallback status.
• A stable JSON result designed for datasets, API calls, and automated workflows.

Input

{
  "monthlyInputTokens": 2000000,
  "monthlyOutputTokens": 500000,
  "liveOnly": false,
  "maxResults": 20
}

Only these four fields are accepted. Token counts must be whole numbers. At least one token count must be greater than zero, and maxResults must be between 1 and 100.

Catalog behavior

The live source is fixed to https://openrouter.ai/api/v1/models; callers cannot replace the URL. If the live catalog is unavailable or invalid, the Actor uses a bundled last-known-good snapshot dated 2026-09-18 unless liveOnly=true. Every result says whether live or fallback data was used and includes its effective timestamp.

Pricing

The price is US$0.10 per successfully stored result. Platform usage is included rather than passed through separately. A valid run produces one dataset item and one cost-scan-result event. Invalid input or a liveOnly catalog failure produces neither.

Privacy and permissions

Submit aggregate token counts only. Do not submit prompts, completions, documents, URLs, file paths, credentials, API keys, customer records, or identifiers. The schema and runtime both reject unknown fields.

The Actor uses Limited permissions, 256 MiB memory, no proxy, no Standby mode, no API key, and no account access. It makes outbound HTTPS requests only to the fixed public OpenRouter catalog. Apify may retain run input, logs, and dataset items under its platform and account-plan policies; no no-store behavior is promised.

Estimate boundary

The ranking applies base prompt and completion token rates only. It excludes request fees, caching, media, web search, reasoning-token differences, tiered or conditional overrides, discounts, retries, credits, taxes, platform fees, and actual provider routing. Provider names come from the OpenRouter model-ID namespace and do not guarantee a serving endpoint.

This is an informational estimate, not a purchasing, architecture, availability, security, or investment recommendation. Verify current model and checkout terms before deciding.