What each route resolves to

Kimi route1 credit
kimi-k3

Moonshot AI

Ultra-long context. Whole documents, repository-scale reading.

Up to 200 messages/month on a standard pass

Claude route2 credits
claude-sonnet-5

Anthropic

Long-form reasoning, writing, code review.

Up to 100 messages/month on a standard pass

Gemini route2 credits
gemini-3.6-flash

Google DeepMind

Fast responses, everyday questions, summarisation.

Up to 100 messages/month on a standard pass

GPT route3 credits
gpt-5.6

OpenAI

Flagship general reasoning.

Up to 66 messages/month on a standard pass

Verify it yourself

Open @LibertiesPassBot, pick any route, and ask:

The answer should match the identifier in the table above. Every route carries a system instruction telling the model to identify itself truthfully — because a gateway that lies about what's under the hood is worth nothing.

Why the weights differ

Credits are a single monthly pool shared across all four routes. A message costs 1, 2 or 3 credits depending on how expensive that engine is for us to run. Cheaper engines stretch further; the flagship costs more. Nothing is hidden in the difference — it simply tracks the bill we pay.

A standard pass carries 200 credits per month. A Legendary pass carries 1,000. The 111 Legendary IDs are scattered across the collection; the list was fixed by a sha256 commitment before the mint. Credits reset on the first of each month, UTC.

When routes change

Providers release, rename and retire models on their own schedule. When a route moves to a newer model, we update this page and announce it in our channel before it takes effect.

What stays fixed: four routes, four independent providers, and a capable current-generation model behind each one. What changes: the specific identifier, as the field moves.

No affiliation

LIBERTIES is independent and not affiliated with, endorsed by, or sponsored by any provider listed here. Names are used solely to state accurately which engine answers which route. All trademarks belong to their respective owners.