Routing
Resolve a model selector into an ordered provider chain, define fallback, and inspect the result.
Routing turns the request's model into an ordered chain of concrete provider endpoints. Most workflows need only a direct selector or a named fallback chain.
Model selector forms
The separator carries routing meaning:
| Form | Meaning | Example |
|---|---|---|
provider/model | A stable logical model id from the registry; every eligible provider that declares it can join the route | anthropic/claude-opus-4.8 |
provider:model | A direct local route through one configured provider | anthropic:anthropic/claude-opus-4.8 |
Prefer the slash form in application code. Use the colon form when the provider itself is part of the decision, such as a subscription-backed account or a specific custom endpoint.
Resolution order
BitRouter resolves a request in four stages:
- Apply a known
@presetand optional:variant. - Treat
provider:modelas an explicit provider route. - Resolve a matching entry under
modelsas a virtual model. - Otherwise, build a cascade from active providers that declare the bare model.
Subscription-backed providers are explicit-route-only. A bare model never silently consumes a personal Claude or Codex subscription.
| Selector | Use it for |
|---|---|
openai:openai/gpt-5.4 | Pin one provider and logical model |
openai/gpt-5.4 | Route a logical model through eligible providers |
coding | Use a named chain from models |
@fast | Apply a reusable preset |
@fast:cost | Apply a preset, then a known variant |
bitrouter/auto | Use the routing policy bound to the auto preset |
Provider selection
For a bare model, BitRouter filters active providers by only, ignore, required tags, capabilities, and protocol compatibility. It then orders candidates by explicit provider priority or the registry's provider-class priority, followed by provider ID.
Presets and variants accept three sort values:
| Value | alpha.31 behavior |
|---|---|
alphabetical | Provider priority, then provider ID |
cost | Accepted, but currently uses the same deterministic fallback order |
latency | Accepted, but currently uses the same deterministic fallback order |
Cost- and latency-based live scoring are not active in alpha.31. Do not describe :cost or :latency as measured optimization until a release provides the corresponding metrics-backed recommender.
Use routing.only, routing.ignore, or provider priority when the order must be explicit today.
Fallback
Define an ordered chain with a virtual model:
models:
coding:
strategy: priority
endpoints:
- provider: openai
service_id: gpt-5
- provider: anthropic
service_id: claude-sonnet-4-6A request for coding tries endpoints in YAML order. BitRouter advances after retryable upstream failures: 5xx, 408, 429, transport or timeout failures, invalid upstream responses, and exhausted provider credit. Other client-side 4xx errors fail immediately.
Use strategy: cascade only when the endpoints are interchangeable and routing preferences may reorder or filter them. In alpha.31, cost and latency sorts still use the deterministic priority/provider-ID fallback described above.
Presets
Presets put model substitution, prompt defaults, parameters, policy binding, and routing filters behind a short name:
presets:
fast:
model: coding
system_prompt: "Be concise."
params:
temperature: 0.2
routing:
only: [openai, anthropic]Call it as @fast. Request fields win over preset defaults, so a preset provides a stable baseline without taking control away from the caller.
bitrouter/auto is the public spelling of the auto preset and requires that preset to be bound to a routing policy. Create the binding with bro policy init rather than hand-authoring a partial lock file.
Variants
A variant changes routing preferences only:
variants:
private:
routing:
require_tags: [private]
primary:
routing:
only: [openai]Append a known variant to a model or preset, such as coding:private or @fast:primary. An unknown suffix is not removed; it remains part of the model selector.
Variants do not grant access, bypass guardrails, or change virtual-key policy.
Request capabilities
Structured outputs and other request capabilities filter the route to compatible provider/model targets before execution. See Structured outputs for the cross-protocol request shapes.
Inspect and verify
bro config validate -c bitrouter.yaml
bro models --provider openai
bro route coding -c bitrouter.yaml
bro route @fast:primary -c bitrouter.yamlbro route uses the running daemon when available and otherwise resolves from the selected config file. It is the authoritative way to review the chain before sending traffic.
How is this guide?