#1 Persistent memory for AI coding agents based on real-world benchmarks
TypeScript agentmemoryagentsaiclaudeclaudecodecodexcopilotcursorgenaiharnesshermesmemoryompopenclawopencodepi
/**
* Shared embedding-dimension logic for OpenAI-compatible providers.
*
* OpenAI and OpenRouter expose the same underlying embedding models, so they
* share one dimension table and one resolver. OpenRouter namespaces model ids
* (e.g. "openai/text-embedding-3-small"); the lookup strips a leading
* "provider/" prefix so both bare and namespaced keys resolve to the same
* dimensions.
*
* The dimension guard (index.ts) throws on mismatch, so a wrong value here
* breaks every embed call — keep entries accurate. Callers pass the relevant
* env-var name (OPENAI_EMBEDDING_DIMENSIONS / OPENROUTER_EMBEDDING_DIMENSIONS)
* so error messages point at the knob the operator actually set.
*/
const MODEL_DIMENSIONS: Record<string, number> = {
"text-embedding-3-small": 1536,
"text-embedding-3-large": 3072,
"text-embedding-ada-002": 1536,
};
const DEFAULT_DIMENSIONS = 1536;
function lookupModelDimensions(rawModel: string): number | undefined {
// OpenRouter variant suffixes (":floor", ":nitro") pick a provider, not a
// different model, so the vectors keep the base model's width.
const model = rawModel.replace(/:[^/:]+$/, "");
if (Object.hasOwn(MODEL_DIMENSIONS, model)) return MODEL_DIMENSIONS[model];
const slash = model.indexOf("/");
if (slash === -1) return undefined;
const bare = model.slice(slash + 1);
return Object.hasOwn(MODEL_DIMENSIONS, bare) ? MODEL_DIMENSIONS[bare] : undefined;
}
export function resolveDimensions(
model: string,
override: string | undefined,
envName: string,
): number {
if (override !== undefined && override.trim().length > 0) {
const parsed = parseInt(override, 10);
if (!Number.isFinite(parsed) || parsed <= 0) {
throw new Error(
`${envName} must be a positive integer, got: ${override}`,
);
}
return parsed;
}
return lookupModelDimensions(model) ?? DEFAULT_DIMENSIONS;
}
export { MODEL_DIMENSIONS, DEFAULT_DIMENSIONS };