feat(models): strong/cheap/vision model slots + multi-provider engine federation
Model slots wave (2026-09-20), chunk B:
- settings: new `models` section {strong, cheap, vision} slots (each
field independent, same shape as a role override); ROLE_SLOT maps all
nine roles to tiers (strong=prober+teacher, cheap=researcher/
demoBuilder/extractor/auditor/labeler, vision=visualMaker+pdfReader);
resolution chain per role: override → slot → factory pin.
- config: ModelSpec gains provider (all pins openrouter, the BYOK
default); MODEL_PINS in stores.ts now spans all nine roles (the strong
slot's factory rep IS the prober pin).
- engine: piModels() registers EVERY builtin provider catalog;
piEngineForSpec(resolved spec) is the new engine factory (pi-ai
resolves per-provider keys from env itself); piEngineFor stays as the
config-pin fallback. chat-handler resolves the conversation chain
(chat override → strong slot → pin) and injects engineFor into the job
runner, closing the old gap where per-role overrides never reached
job engines.
- /api/models: models federate ONLY keyed providers (providers surface
with keyPresent, env-wins, fresh per call); slots (resolved tier pins)
join rolePins; catalogDefaultsFor inherits the strong slot; the models
settings section gets the same catalog gate as chat.
- pricing: PRICES/CONTEXT_LIMITS uniformly keyed provider:id
(regenerated from ALL pi-ai catalogs, 1354 keys); usage stamps
(TurnUsage, JobUsage, extraction stages) switch to provider:id —
owner decision: no legacy compat, old stamps render unpriced.
This commit is contained in:
@@ -57,6 +57,51 @@ export interface RoleModelOverride {
|
||||
thinkingLevel?: ThinkingLevel;
|
||||
}
|
||||
|
||||
// ─── Model slots (model slots wave, 2026-09-20) ──────────────────────────────
|
||||
|
||||
/** The capability tiers a user can pin models into (owner-settled mapping):
|
||||
* strong = prober + teacher (also the per-session default-teacher
|
||||
* starting point); cheap = researcher, demo-builder, extractor, auditor,
|
||||
* labeler; vision = visual-maker + pdf-reader. Resolution per role:
|
||||
* ROLE OVERRIDE → slot → factory pin. */
|
||||
export const MODEL_SLOTS = ["strong", "cheap", "vision"] as const;
|
||||
export type ModelSlotName = (typeof MODEL_SLOTS)[number];
|
||||
|
||||
/** One slot's override — same per-field independence as a role override:
|
||||
* absent model keeps the tier's current model, absent thinkingLevel keeps
|
||||
* its thinking budget. */
|
||||
export interface ModelSlot {
|
||||
/** The model swap: both coordinates or neither. */
|
||||
model?: { provider: string; id: string };
|
||||
/** The thinking budget swap. */
|
||||
thinkingLevel?: ThinkingLevel;
|
||||
}
|
||||
|
||||
/** The `models` settings section (owner decision 2026-09-20: a dedicated
|
||||
* section — slots are engine-level choices, not one session's). All slots
|
||||
* optional: an absent slot means every role of its tier runs whatever the
|
||||
* deeper tiers of the chain resolve to (per-role override → factory pin). */
|
||||
export interface ModelsSettings {
|
||||
strong?: ModelSlot;
|
||||
cheap?: ModelSlot;
|
||||
vision?: ModelSlot;
|
||||
}
|
||||
|
||||
/** Which slot a role draws from (model slots wave, 2026-09-20). Includes the
|
||||
* conversation pair and pdfReader (not BackgroundModelRoles — the slot
|
||||
* tier spans every role; only background roles surface per-role overrides). */
|
||||
export const ROLE_SLOT: Readonly<Record<string, ModelSlotName>> = {
|
||||
prober: "strong",
|
||||
teacher: "strong",
|
||||
researcher: "cheap",
|
||||
demoBuilder: "cheap",
|
||||
extractor: "cheap",
|
||||
auditor: "cheap",
|
||||
labeler: "cheap",
|
||||
visualMaker: "vision",
|
||||
pdfReader: "vision",
|
||||
};
|
||||
|
||||
/** General settings — app-level behavior. Version DISPLAY and plans login
|
||||
* live outside the record (package.json / auth.json) by owner decision. */
|
||||
export interface GeneralSettings {
|
||||
@@ -140,6 +185,8 @@ export interface UserSettings {
|
||||
* controls. NOT shown in the settings panel (owner decision). */
|
||||
chat?: ChatSettings;
|
||||
agent?: AgentSettings;
|
||||
/** The strong/cheap/vision model slots (model slots wave, 2026-09-20). */
|
||||
models?: ModelsSettings;
|
||||
appearance?: AppearanceSettings;
|
||||
graph?: GraphSettings;
|
||||
[key: string]: unknown;
|
||||
@@ -244,6 +291,18 @@ function sanitizeRoleOverride(raw: unknown): RoleModelOverride | undefined {
|
||||
return Object.keys(out).length ? out : undefined;
|
||||
}
|
||||
|
||||
/** The models section (model slots wave, 2026-09-20): each slot keeps
|
||||
* only well-formed parts — the same shape rules as a role override, so a
|
||||
* stale client can't park junk in a slot. */
|
||||
function sanitizeModels(raw: Record<string, unknown>): ModelsSettings {
|
||||
const out: ModelsSettings = {};
|
||||
for (const slot of MODEL_SLOTS) {
|
||||
const cleaned = sanitizeRoleOverride(raw[slot]);
|
||||
if (cleaned) out[slot] = cleaned;
|
||||
}
|
||||
return out;
|
||||
}
|
||||
|
||||
function sanitizeBooleanMap(raw: unknown): Record<string, boolean> | undefined {
|
||||
if (raw === null || typeof raw !== "object" || Array.isArray(raw)) return undefined;
|
||||
const out: Record<string, boolean> = {};
|
||||
@@ -348,6 +407,10 @@ export function sanitizeUserSettings(raw: unknown): UserSettings | null {
|
||||
if (r.appearance !== null && typeof r.appearance === "object" && !Array.isArray(r.appearance)) {
|
||||
out.appearance = sanitizeAppearance(r.appearance as Record<string, unknown>);
|
||||
}
|
||||
if (r.models !== null && typeof r.models === "object" && !Array.isArray(r.models)) {
|
||||
const models = sanitizeModels(r.models as Record<string, unknown>);
|
||||
if (Object.keys(models).length) out.models = models;
|
||||
}
|
||||
if (r.graph !== null && typeof r.graph === "object" && !Array.isArray(r.graph)) {
|
||||
out.graph = sanitizeGraph(r.graph as Record<string, unknown>);
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user