> ## Documentation Index > Fetch the complete documentation index at: https://docs.together.ai/llms.txt > Use this file to discover all available pages before exploring further. # Prepare a fine-tuned model for FP4 inference > Merges a fine-tuned adapter into its base model and prepares the result for FP4 inference. The adapter is resolved when the request is made, so a request naming a model that does not exist in the project, is not an adapter, was not produced by fine-tuning, has no base model to merge into, or uses an unsupported base model is rejected with 400 rather than accepted and failed later. The run is priced before it is created, and a project that cannot cover it is answered 402. ## OpenAPI ````yaml openapi.yaml POST /shaping/prepare-for-fp4-inference openapi: 3.1.0 info: title: Together APIs description: The Together REST API. See https://docs.together.ai for more details. version: 2.0.0 termsOfService: https://www.together.ai/terms-of-service contact: name: Together Support url: https://www.together.ai/contact license: name: MIT url: https://github.com/togethercomputer/openapi/blob/main/LICENSE servers: - url: https://api.together.ai/v1 description: Default environment for APIs - url: https://api-inference.together.ai/v2 description: Optimized environment for inference security: - bearerAuth: [] paths: /shaping/prepare-for-fp4-inference: post: tags: - Shaping summary: Prepare a fine-tuned model for FP4 inference description: > Merges a fine-tuned adapter into its base model and prepares the result for FP4 inference. The adapter is resolved when the request is made, so a request naming a model that does not exist in the project, is not an adapter, was not produced by fine-tuning, has no base model to merge into, or uses an unsupported base model is rejected with 400 rather than accepted and failed later. The run is priced before it is created, and a project that cannot cover it is answered 402. operationId: createQuantizationJob requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/QuantizationRequest' responses: '200': description: Quantization job created successfully. content: application/json: schema: $ref: '#/components/schemas/QuantizationJob' '400': description: Invalid request parameters. content: application/json: schema: $ref: '#/components/schemas/ErrorData' '401': description: Missing or invalid API key. content: application/json: schema: $ref: '#/components/schemas/ErrorData' '402': description: The project cannot cover the estimated price of the run. content: application/json: schema: $ref: '#/components/schemas/ErrorData' '403': description: The project is blocked and cannot start a run. content: application/json: schema: $ref: '#/components/schemas/ErrorData' '500': description: Internal server error. content: application/json: schema: $ref: '#/components/schemas/ErrorData' '503': description: Preparing models for FP4 inference is not available. content: application/json: schema: $ref: '#/components/schemas/ErrorData' components: schemas: QuantizationRequest: type: object description: Request body for preparing an adapter for FP4 inference. required: - inputs properties: inputs: $ref: '#/components/schemas/QuantizationPipelineInputs' description: Adapter inputs to prepare. QuantizationJob: description: Quantization job that prepares an adapter for FP4 inference. allOf: - $ref: '#/components/schemas/BaseShapingJob' - type: object description: Quantization job that prepares an adapter for FP4 inference. required: - params - events properties: params: $ref: '#/components/schemas/QuantizationPipelineSpec' description: Job specification. results: $ref: '#/components/schemas/QuantizationResults' description: Model artifacts produced by the job. events: type: array description: Events emitted by the job. items: $ref: '#/components/schemas/QuantizationEvent' ErrorData: type: object required: - error properties: error: type: object properties: message: type: string nullable: false type: type: string nullable: false param: type: string nullable: true default: null code: type: string nullable: true default: null required: - type - message QuantizationPipelineInputs: type: object description: Adapter identifiers to prepare. required: - adapter_object_id properties: adapter_object_id: type: string pattern: ^ml_.+ description: Model object ID of the adapter to prepare. adapter_revision_id: type: string pattern: ^rv_.+ description: >- Adapter revision ID to prepare. Omit to use the adapter's current revision. calibration_file_id: type: string pattern: ^file-.+ description: > Conversation dataset file ID to use for calibration. Upload it first with `POST /v1/files` using purpose `calibration` and file type `jsonl`. Each JSONL row must contain a `messages` array and may contain a `tools` array. When provided, the job draws half of the calibration corpus from this file and half from a general-text corpus. When omitted, the job calibrates on the general-text corpus only. The request is rejected unless the file exists in the request project, has finished uploading, has passed files API validation as a conversation dataset, is not empty, and is at most 64 MiB. A `fine-tune` purpose file is also accepted so a training file can be reused. example: file-5f1b0c7a-2d19-4a3e-9c60-8ab41c2f7d35 example: adapter_object_id: ml_CeG3fF6pyEViU8dE8pk7R adapter_revision_id: rv_CeG3faPBj2ABwbAaHPTmx calibration_file_id: file-5f1b0c7a-2d19-4a3e-9c60-8ab41c2f7d35 BaseShapingJob: type: object description: Public fields common to shaping jobs. required: - id - job_type - status - created_at - updated_at - user_id properties: id: type: string pattern: ^shp-quant-[a-f0-9-]+$ description: Unique shaping job ID. job_type: $ref: '#/components/schemas/ShapingJobType' description: Type of shaping job. status: $ref: '#/components/schemas/ShapingJobStatus' description: Current job status. created_at: type: string format: date-time description: Time when the job was created. updated_at: type: string format: date-time description: Time when the job was last updated. user_id: type: string description: User that created the job. QuantizationPipelineSpec: type: object description: >- Stored job specification containing the request and resolved configuration. required: - inputs - configuration properties: inputs: $ref: '#/components/schemas/QuantizationPipelineInputs' description: Adapter inputs requested by the caller. configuration: $ref: '#/components/schemas/QuantizationPipelineConfiguration' description: Server-resolved configuration for the run. QuantizationResults: type: object description: Model artifact produced by a completed quantization job. properties: model_object_id: type: string description: Model object ID registered by the job. model_revision_id: type: string description: Model revision ID registered by the job. QuantizationEvent: description: Event emitted by a quantization job. allOf: - $ref: '#/components/schemas/BaseShapingEvent' - type: object description: Event emitted by a quantization job. properties: type: type: string description: Quantization event type. enum: - SHAPING_ERROR - SHAPING_USER_ERROR - QUANTIZATION_JOB_START - QUANTIZATION_MODEL_DOWNLOAD_COMPLETE - QUANTIZATION_MERGE_COMPLETE - QUANTIZATION_QUANTIZE_PROCESS_START - QUANTIZATION_QUANTIZE_PROCESS_COMPLETE - QUANTIZATION_MODEL_UPLOAD_COMPLETE - QUANTIZATION_JOB_COMPLETE result: $ref: '#/components/schemas/QuantizationResults' description: Result attached to completion events. ShapingJobType: type: string description: Kind of shaping job. enum: - quantization ShapingJobStatus: type: string description: Lifecycle status of a shaping job. enum: - pending - queued - running - completed - cancelled - cancel_requested - error - user_error QuantizationPipelineConfiguration: type: object description: Server-resolved configuration for a quantization job. readOnly: true required: - user_id - adapter_project_id - adapter_revision_id - adapter_model_name - base_object_id - base_revision_id - base_model_name properties: user_id: type: string description: User whose scope is used to fetch the adapter weights. example: user_CKfewYkQBpaEmN7wuQS3W adapter_project_id: type: string description: Project that owns the adapter. example: proj_CczF9KYFantA84sNYDTJd adapter_revision_id: type: string description: Adapter revision selected for preparation. example: rv_CeERyXMBQCJTRC1yqGTpk adapter_model_name: type: string description: Qualified adapter model name. example: ft-qa-artifacts/glm-5.3-e2e-test-adapter base_object_id: type: string description: Model object ID of the adapter's base model. example: ml_CbJ9yCnij7A47b1xkpioB base_revision_id: type: string description: Revision ID of the adapter's base model. example: rv_CbJ9yNrws93VQrZum1fTE base_model_name: type: string description: Qualified name of the base model. example: zai-org/GLM-5.3-NVFP4 BaseShapingEvent: type: object description: Public fields common to shaping events. required: - object - created_at - level - message - type properties: object: type: string const: shaping description: Object type, always `shaping`. created_at: type: string format: date-time description: Time when the event was created. level: $ref: '#/components/schemas/EventLevel' description: Event severity level. message: type: string description: Human-readable event message. type: type: string description: Event type identifier. hash: type: string description: Event hash used for deduplication. EventLevel: type: string description: Severity level for a shaping event. enum: - Info - Warning - Error securitySchemes: bearerAuth: type: http scheme: bearer x-bearer-format: bearer x-default: default ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.