AI Potluck
Product / UX / UI & API

OpenRouter

OpenRouter

Commercial API gateway exposing models from many providers behind one OpenAI-compatible endpoint, with automatic failover, routing and unified billing across them. It runs as a hosted service and is a routing gateway rather than a chat interface.

The homepage figures the adoption band was originally scored on have since moved, and the band was re-derived against the current ones. Verified 2026-08-13 via the OpenRouter homepage.

Openness

1 high confidence
1.0
license
proprietary(hosted gateway)
source
closed(SDK present, core service closed)
self-host
none

A proprietary hosted routing gateway with no published source and no self-hosting: the closed counterpart to open gateways such as LiteLLM.

  • https://openrouter.ai/ recorded 2026-08-13

    hosted unified API gateway across 400+ models/60+ providers; proprietary service (no self-host)

Adoption

5 high confidence
5.0

The quantity banded here is the vendor's own "global users" tile on the OpenRouter homepage, which reads 10M+ and clears the >10M threshold the top band needs. The same tiles read 200T+ monthly tokens, 80+ providers and 500+ models. Token volume is corroborated externally at roughly 25T tokens a week, a run-rate near 1.5 quadrillion a year, per Menlo, and by TechCrunch on the $1.3B valuation and $113M raise. Two caveats on the instrument: the unit is a vendor-reported user count rather than the package downloads this scale is usually denominated in, though the label is the top band either way; and OpenRouter does not say whether "global users" is an active or a cumulative count, which is why this stays a usage-volume band rather than an active-user one, where a cumulative total would not be admissible. The 200T+ monthly tokens is the corroborating measure and is unambiguously a volume.

Capability

5 high confidence
5.0

Frontier-defining for the gateway slot: the broadest provider coverage plus routing and fallback at the largest measured token scale, and the closed benchmark that LiteLLM approaches on the open side. The homepage stat tiles now read 200T+ monthly tokens, 400+ models and 70+ providers, so the breadth that made this the gateway frontier is if anything wider.

  • https://openrouter.ai/ recorded 2026-08-13

    400+ models, 60+ providers, routing/fallback, cost tracking, OpenAI-compat

Verified 2026-08-13