> ## Documentation Index > Fetch the complete documentation index at: https://docs.sqwish.ai/llms.txt > Use this file to discover all available pages before exploring further. # Keep a model warm > Keep a model loaded between requests on its own, by its number such as router@4, or stop. Its own switch must be on: 409 model_off otherwise, and switching it off stops keep warm too. Switching on loads and probes it first, then holds its first hour, at most `max_price_usd`. A model kept warm through its family as well takes one warm place and is billed once. `switch_off` stops keep warm on families or models in the same step and leaves their switches on, so a request at the limit can swap. Past the account's limit, 409 warm_limit lists the models still kept warm after that, least recently used first (`used_at`), each with the keep warms that hold it: its own (`keep_warm`) and its families' (`families`). A size with no kept-warm place left answers 503 warm_capacity, unless `switch_off` stops a model of that size. With `adapter`, only while the number names the model the caller saw: 409 model_changed otherwise, before anything loads. ## OpenAPI ````yaml /openapi.json post /v1/models/{model}/keep-warm openapi: 3.1.0 info: title: DecisionOne workbench description: >- Native decisions, versioned datasets, adaptation and shared-base deployment. Authenticate with `Authorization: Bearer d1_sk_...` when accounts are on. A short guide for coding agents is at /llms.txt. Hosted account APIs accept a Bearer API key or a browser session. An API key reads its account but can't change it: keys, billing and settings change in the console. Browser mutations require the configured Origin and X-CSRF-Token from /auth/session. Account recovery and history remain available to blocked accounts. Waitlisted accounts can read their account summary and model catalogue and request early access; private member APIs require admission. Money fields ending in _usd are exact decimal strings; never interpret them as integer cents. Authenticated customer reads and existing-operation replays share read_rpm, and other customer writes except decisions and estimates share write_rpm. Exhausted budgets return 429 with Retry-After; unavailable admission storage returns 503. version: 0.1.0 servers: - url: https://console.sqwish.ai security: [] tags: - name: Decisions description: Probabilities for questions with fixed outcomes. - name: Models description: Base models, adapters and named models with versions. - name: Families description: >- Lines of models trained on one dataset, and the switches that let them answer calls. - name: Datasets description: >- Validated training examples. A dataset grows in versions, and a version never changes. - name: Evaluations description: Reusable model results on a dataset's fixed evaluation rows. - name: Fine-tuning description: Training jobs, their events and held-out results. - name: Prompt tuning description: Better wording for a decision, tested on your rows. Words, not weights. - name: Labeling description: Teacher labels for real queries, reviewed by you. - name: Accounts description: Email OTP sign-in, revocable sessions, security and API keys. - name: Billing description: Exact USD balances, purchases, automatic top-ups and referrals. - name: Integrations description: Endpoints other tools call, such as Claude Code. - name: System description: Health and discovery. paths: /v1/models/{model}/keep-warm: post: tags: - Models summary: Keep a model warm description: >- Keep a model loaded between requests on its own, by its number such as router@4, or stop. Its own switch must be on: 409 model_off otherwise, and switching it off stops keep warm too. Switching on loads and probes it first, then holds its first hour, at most `max_price_usd`. A model kept warm through its family as well takes one warm place and is billed once. `switch_off` stops keep warm on families or models in the same step and leaves their switches on, so a request at the limit can swap. Past the account's limit, 409 warm_limit lists the models still kept warm after that, least recently used first (`used_at`), each with the keep warms that hold it: its own (`keep_warm`) and its families' (`families`). A size with no kept-warm place left answers 503 warm_capacity, unless `switch_off` stops a model of that size. With `adapter`, only while the number names the model the caller saw: 409 model_changed otherwise, before anything loads. operationId: keepModelWarm parameters: - name: model in: path required: true schema: type: string title: Model requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/ModelKeepWarm' responses: '200': description: Successful Response content: application/json: schema: properties: id: type: string title: Id status: type: string title: Status evaluation_key: type: string title: Evaluation Key name: type: string title: Name description: type: string title: Description family: type: string title: Family size_order: type: integer title: Size Order kind: type: string title: Kind lora: anyOf: - type: string - type: 'null' title: Lora metrics: anyOf: - additionalProperties: true type: object - type: 'null' title: Metrics max_prompt_tokens: type: integer title: Max Prompt Tokens max_questions: type: integer title: Max Questions calibration: type: string title: Calibration availability_reason: type: string title: Availability Reason description: >- Why it can't answer now: no_engine (its pool has no engine right now), not_served (this server doesn't serve it), or, for an adapter, base_mismatch, engine_unconfigured or engine_changed. available: type: boolean title: Available access_requested: type: boolean title: Access Requested adapter_capacity: anyOf: - $ref: '#/components/schemas/AdapterCapacity' - type: 'null' created_at: type: number title: Created At deleted_at: type: number title: Deleted At description: >- When its owner deleted it. A deleted model answers and runs nothing. discarded_at: type: number title: Discarded At description: >- When a rollback of its family discarded it. A discarded model answers and runs nothing. on_at: anyOf: - type: number - type: 'null' title: On At description: >- When its own switch went on: it answers calls to its number, such as router@4. Null while it is off. keep_warm: type: boolean title: Keep Warm description: >- Whether it is kept warm on its own switch. A family's keep warm doesn't set it: see keep_warm_window. keep_warm_window: anyOf: - $ref: '#/components/schemas/KeepWarmWindow' - type: 'null' description: >- The hour of keep warm being billed for it, through its family or its own switch, or null when nothing is billed. keep_warm_stopped_at: anyOf: - type: number - type: 'null' title: Keep Warm Stopped At description: >- When its own keep warm switched off because the next hour couldn't be reserved, while it stays off. Otherwise null. used_at: anyOf: - type: number - type: 'null' title: Used At description: >- Its idle clock: the latest of its last production call by an API key that named it, its last evaluation that named it, and when it was switched on. switches_off_at: anyOf: - type: number - type: 'null' title: Switches Off At description: >- When idling switches it off unless it's used. Null while it is off or kept warm on its own switch, and when idling never switches anything off. A family's is also null while it serves no model. off_reason: anyOf: - type: string enum: - owner - idle - type: 'null' title: Off Reason description: >- Why it is off: owner when it was switched off by hand, idle when nothing used it for too long. Null otherwise. off_at: anyOf: - type: number - type: 'null' title: Off At description: When it went off, with off_reason. Null otherwise. job_id: type: string title: Job Id model_name: anyOf: - type: string - type: 'null' title: Model Name version: type: integer title: Version parent: anyOf: - type: string - type: 'null' title: Parent residency: $ref: '#/components/schemas/Residency' state: anyOf: - type: string enum: - loaded - loading - asleep - type: 'null' title: State decision_price_nano: anyOf: - type: string - type: 'null' title: Decision Price Nano description: >- Exact integer string of nano-USD per billed input token from the current billing policy. Zero means free; null means unpriced. base: type: string title: Base description: 'An adapter''s base model id: the base whose prices it pays.' evaluation_set: anyOf: - $ref: '#/components/schemas/EvaluationSet' - type: 'null' request_latency: $ref: '#/components/schemas/RequestLatency' type: object required: - id - status - evaluation_key title: Model description: A base, an early-access catalogue entry, or a trained adapter. 4XX: description: >- Client error. On a 422, `error.details` lists each field that failed. content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' 5XX: description: Server error. A 503 names what is not running on this server. content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' security: - BearerAuth: [] components: schemas: ModelKeepWarm: properties: enabled: type: boolean title: Enabled max_price_usd: anyOf: - type: string pattern: ^\d{1,9}(\.\d{1,9})?$ - type: 'null' title: Max Price Usd description: >- The most you agree to pay for the first hour, which switching keep warm on holds, in USD, such as the price you were shown. If it costs more when it is admitted, it is refused with price_changed and nothing is held. adapter: type: string maxLength: 140 title: Adapter description: >- The id of the model the caller saw: refused unless the model called is still that one. switch_off: items: type: string type: array maxItems: 20 title: Switch Off description: >- Families, such as router, and models, such as router@2, whose keep warm stops in the same step, to make room when the account is at its limit (warm_limit) or the model's size has no kept-warm place left (warm_capacity). Their switches stay on. additionalProperties: false type: object required: - enabled title: ModelKeepWarm AdapterCapacity: properties: max_rank: type: integer title: Max Rank slots: anyOf: - type: integer - type: 'null' title: Slots type: object required: - max_rank - slots title: AdapterCapacity KeepWarmWindow: properties: started_at: type: number title: Started At ends_at: type: number title: Ends At description: When this hour's hold ends and the next is reserved. reserved_usd: type: string title: Reserved Usd description: >- The hour's fee held, as exact USD. The minutes used are charged when the hour rolls over or keep warm is switched off. type: object required: - started_at - ends_at - reserved_usd title: KeepWarmWindow Residency: properties: state: type: string enum: - hot - cold title: State pinned: type: boolean title: Pinned type: object required: - state - pinned title: Residency EvaluationSet: properties: dataset: $ref: '#/components/schemas/EvaluationDataset' training_examples: anyOf: - type: integer - type: 'null' title: Training Examples validation_examples: anyOf: - type: integer - type: 'null' title: Validation Examples remembered_examples: anyOf: - type: integer - type: 'null' title: Remembered Examples type: object required: - dataset - training_examples - validation_examples - remembered_examples title: EvaluationSet RequestLatency: properties: median_ms: anyOf: - type: number - type: 'null' title: Median Ms samples: type: integer title: Samples median_is_lower_bound: type: boolean title: Median Is Lower Bound start: type: string title: Start end: type: string title: End type: object required: - median_ms - samples - median_is_lower_bound - start - end title: RequestLatency ErrorResponse: description: >- The body of every error. app.failure() builds it; this model documents it. properties: error: $ref: '#/components/schemas/ErrorBody' required: - error title: ErrorResponse type: object EvaluationDataset: properties: id: type: string title: Id name: anyOf: - type: string - type: 'null' title: Name archived_at: anyOf: - type: number - type: 'null' title: Archived At deleted: type: boolean title: Deleted description: True once you deleted it. type: object required: - id - name - archived_at - deleted title: EvaluationDataset ErrorBody: properties: code: description: A stable code to branch on, such as invalid_request. title: Code type: string message: description: What went wrong and what to change, in plain words. title: Message type: string retryable: description: >- True when another attempt may succeed; it does not prove that no work was done. Respect Retry-After when present. Only operations that support Idempotency-Key replay a prior result, including /v1/decide for 24 hours. title: Retryable type: boolean request_id: description: Also sent as X-Request-ID. Quote it in bug reports. title: Request Id type: string details: anyOf: - items: $ref: '#/components/schemas/ErrorDetail' type: array - type: 'null' default: null description: On validation errors, one entry for each field that failed. title: Details required: - code - message - retryable - request_id title: ErrorBody type: object ErrorDetail: properties: path: description: Where the problem is, such as ["body", "decisions", 0, "outcomes"]. items: anyOf: - type: string - type: integer title: Path type: array message: title: Message type: string type: anyOf: - type: string - type: 'null' default: null title: Type required: - path - message title: ErrorDetail type: object securitySchemes: BearerAuth: type: http scheme: bearer bearerFormat: d1_sk_... description: Create an API key in the console. Required when accounts are enabled. ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.