Rulvar

Appearance

Rulvar API reference

Rulvar API reference / @rulvar/core / ModelCaps

Type Alias: ModelCaps ​

ts
type ModelCaps = {
  contextWindow: number;
  maxOutputTokens: number;
  minOutputTokensPerTurn?: number;
  pricing?: Pricing;
  promptCaching?: "explicit" | "implicit";
  reasoningEfforts: Effort[];
  structuredOutput: "native" | "forced-tool" | "prompt";
  supportsParallelTools: boolean;
  supportsTemperature: boolean;
};

Defined in: packages/core/src/l0/spi/provider.ts:97

Capability facts the router consumes for tier selection and scrubbing.

Properties ​

contextWindow ​

ts
contextWindow: number;

Defined in: packages/core/src/l0/spi/provider.ts:103

maxOutputTokens ​

ts
maxOutputTokens: number;

Defined in: packages/core/src/l0/spi/provider.ts:104

minOutputTokensPerTurn? ​

ts
optional minOutputTokensPerTurn?: number;

Defined in: packages/core/src/l0/spi/provider.ts:114

The smallest request output cap the provider accepts (the v1.74 experiment review, P0.1): OpenAI's Responses API rejects max_output_tokens below 16, so a dispatch under this floor is a guaranteed 400. The runtime never sends a request output cap below it: a budget last gasp dispatches the floor instead of one token, and a remainder that cannot buy the floor is refused typed before the wire. Absent means one, the historical floor.

pricing? ​

ts
optional pricing?: Pricing;

Defined in: packages/core/src/l0/spi/provider.ts:126

Adapter-reported fallback only; the versioned price table wins.

promptCaching? ​

ts
optional promptCaching?: "explicit" | "implicit";

Defined in: packages/core/src/l0/spi/provider.ts:124

How this model's prompt caching is driven (RV2006). 'explicit' means the adapter compiles ChatRequest.cacheHint into provider cache directives (Anthropic cache_control) and the agent loop's cache policy attaches hints by default; 'implicit' means the provider caches server-side on its own and hints are neither needed nor sent (OpenAI). Absent means unknown: the loop attaches nothing and the wire stays byte identical to pre-RV2006 traffic.

reasoningEfforts ​

ts
reasoningEfforts: Effort[];

Defined in: packages/core/src/l0/spi/provider.ts:102

Canonical efforts this model accepts after mapping.

structuredOutput ​

ts
structuredOutput: "native" | "forced-tool" | "prompt";

Defined in: packages/core/src/l0/spi/provider.ts:98

supportsParallelTools ​

ts
supportsParallelTools: boolean;

Defined in: packages/core/src/l0/spi/provider.ts:100

supportsTemperature ​

ts
supportsTemperature: boolean;

Defined in: packages/core/src/l0/spi/provider.ts:99