Public pricing benchmark

Which AI API Proxies Offer the Biggest Savings on the Latest Models?

Use official input and output prices as a common benchmark. Switch models to find the listed AI API proxies with the greatest observed savings.

Anthropic

  1. Xclis.ai logoXclis.aiSave 97.8%
  2. API-Route logoAPI-RouteSave 94.2%
  3. MuskAI logoMuskAISave 92.0%
  4. APIKey logoAPIKeySave 88.6%
  5. RunAPI logoRunAPISave 88.3%
  6. Yomi API logoYomi APISave 88.0%
  7. WoKeyAI logoWoKeyAISave 85.0%
  8. kukuai logokukuaiSave 82.4%
  9. TeamoRouter logoTeamoRouterSave 76.6%
  10. DeepKey logoDeepKeySave 75.0%

OpenAI

  1. kukuai logokukuaiSave 95.2%
  2. API-Route logoAPI-RouteSave 94.4%
  3. OmniaKey logoOmniaKeySave 93.0%
  4. RunAPI logoRunAPISave 92.6%
  5. TeamoRouter logoTeamoRouterSave 91.7%
  6. WoKeyAI logoWoKeyAISave 91.0%
  7. derouter.ai logoderouter.aiSave 90.5%
  8. OpenModel logoOpenModelSave 90.0%
  9. Poixe AI logoPoixe AISave 50.0%
  10. OpenOx logoOpenOxSave 20.0%

Google

  1. MuskAI logoMuskAISave 88.0%
  2. APIKey logoAPIKeySave 85.7%
  3. kukuai logokukuaiSave 82.7%
  4. RunAPI logoRunAPISave 78.9%
  5. TeamoRouter logoTeamoRouterSave 76.0%

xAI

  1. kukuai logokukuaiSave 96.8%
  2. MuskAI logoMuskAISave 96.0%
  3. API-Route logoAPI-RouteSave 95.3%
  4. DeepKey logoDeepKeySave 94.0%
  5. RunAPI logoRunAPISave 93.0%
  6. APIKey logoAPIKeySave 85.7%
  7. Yomi API logoYomi APISave 84.0%
  8. WoKeyAI logoWoKeyAISave 80.0%
  9. Crazyrouter logoCrazyrouterSave 15.0%
  10. OpenModel logoOpenModelSave 5.0%

DeepSeek

  1. OpenModel logoOpenModelSave 75.0%
  2. TokenMix logoTokenMixSave 69.9%
  3. TokensMind logoTokensMindSave 68.2%
  4. APIKey logoAPIKeySave 63.3%
  5. WoKeyAI logoWoKeyAISave 60.0%
  6. Yomi API logoYomi APISave 55.0%
  7. Crazyrouter logoCrazyrouterSave 47.7%
  8. MuskAI logoMuskAISave 41.1%
  9. RunAPI logoRunAPISave 19.1%

Submit public pricing for review

Share your official site and public pricing for review against the listed models.

Submit a provider

How the price rankings work

Savings are a price reference, not a quality score. Verify the provider's current pricing, model mapping, limits, and terms before paying.

  1. Same-day quote

    Only public quotes captured on the page's snapshot date are included; when a site has multiple variants, the quote with the lowest ratio is used.

  2. Same model

    The comparison uses the official input and output prices for the exact model in the header.

  3. Lower is better

    The larger of the input or output price ratios sets the savings percentage.

AI proxy price rankings: coverage and limitations

Understand which quotes are included and what this price-only comparison cannot tell you.

  • Public quotes only

    The list uses public input and output prices captured on the snapshot date. Private offers, hidden discounts, and quotes we cannot compare are excluded.

  • Up to ten providers per model

    Each model column can show up to ten eligible providers. Some models have fewer entries because fewer providers publish a comparable quote.

  • Price is not quality

    The ranking does not measure latency, uptime, routing quality, privacy, support, or refunds. Check those factors on the provider detail page.

Questions about price rankings

See how prices are compared, which services make the list, and how to read the savings percentage.

What does the savings percentage mean?

It is the difference between the observed proxy ratio and the official price for the exact model. A 90% saving means the higher of the input or output ratios is about 10% of the official price.

Why does each column use a different list of services?

The columns are ranked independently. A service must currently list that exact model and have a comparable quote from the page's snapshot date, so coverage can differ by model.

How is the default model selected?

We follow the maintained model list from newest to oldest and select the first model with an official input and output benchmark plus current provider quotes. Use the model selector beside the name to switch to another model with current comparable data.

Why do some rankings have fewer than ten entries?

The page shows only current, comparable public quotes. We leave missing places unfilled instead of inventing a rank or carrying forward stale pricing.

Does a higher savings percentage mean better service?

No. This is a price reference only. Check model routing, rate limits, uptime, privacy terms, support, refunds, and a small test call before relying on a provider.

How should I use these rankings to choose a proxy?

Select the model you plan to use, compare each proxy's savings percentage and input/output quotes, then open the provider details to verify model mapping, rate limits, billing rules, and terms before testing with a small request.