Support automatically switching provider for similar or same models

[charlieadair]

Before submitting:

• I searched existing issues and did not find a duplicate.
• I am describing a concrete problem or use case, not just a vague idea.

Area

provider orchestration ?

Problem or use case

I have access to the same model from multiple different places. For example: Claude Fable is available from my max subscription AND provided by GitHub copilot (via Opencode) with my education benefits alongside other Anthropic and Codex models as well. Additionally, OpenRouter has options for all of these.

Proposed solution:

It would be nice if there was a way to recognize that these are all the same model underneath and have a way to route to the next cheapest, or prioritize using THIS provider before THAT provider.

I imagine this would work by first grouping similarly named models between providers > configuring priority > showing it as one model within the UI. When you hit a 402 payment required for the next API request, switch to the next provider. This logic could probably extend to switching similarly capable models (Astra to Fable, for instance).

Why this matters

High potential to save money and utilize free and/or cheaper providers first. Also allows you to automatically switch to a different provider and use the specified model even when the first endpoint is down or busy.

Having multiple providers, particularly in OpenCode also causes confusion on which opencode/anthropic/X I am hitting. Whether its from GitHub copilot, Amazon Bedrock, etc.

Smallest useful scope

Once I have taken advantage of free usage limits within GitHub copilot's Fable endpoint (via OpenCode) and hit a 402 -> auto switch to my claude subscription. Once I am out of fable credits on my claude subscription -> use Amazon Bedrock via openrouter.

Alternatives considered:

From my understanding, #2829 adds the ability to do cross-provider orchestration. This could just be a definition for the agent to use from the specific agent markdown definition. "If this provider gives you problems -> use this one." This isn't a great solution however because it still crowds your model selector with identical names.

Risks or tradeoffs

In some situations, taking advantage of prompt cache-ing is going to be cheaper than switching to a different provider.

Examples or references

Idk what this means, sorry.

Contribution

• I would be open to helping implement this.

I would love to get the opportunity to contribute to an open source project like this. Thanks.