> ## Documentation Index > Fetch the complete documentation index at: https://docs.veri.studio/llms.txt > Use this file to discover all available pages before exploring further. # Roll back an adapter > Point the alias back at the previously served adapter model (previous_model_id) under a fresh version. Status returns to 'attaching' until every serving replica reports the rollback target loaded: usually immediate when the old weights are still resident, otherwise after a re-download. ## OpenAPI ````yaml /api-reference/openapi.json post /v1/deployments/{deployment_id}/adapters/{serve_name}/rollback openapi: 3.1.0 info: title: Veri API description: >- REST API for the Veri RL post-training platform. All requests require a Bearer API key (`vk_` prefix). license: name: '' version: 0.1.0 servers: - url: https://api.veri.studio description: Production security: [] tags: - name: Training jobs description: Create, monitor, and manage training jobs. - name: Datasets description: Upload and connect training datasets. - name: Deployments description: Serve trained models and run inference. - name: Volumes description: Persistent file storage mounted into jobs. - name: Models description: Custom model registry deployments serve from. - name: Regions description: Discover available launch regions. - name: GPU description: Live GPU availability by provider and region. - name: Code artifacts description: Upload custom training script bundles. - name: Billing description: Credit balance and transaction history. - name: API keys description: Create and revoke API keys. - name: Account description: The authenticated caller's identity. - name: Settings description: Account-level integrations (Weights & Biases). - name: Metrics description: Prometheus metrics export for your own observability stack. - name: Evaluators description: 'Evaluators: versioned scoring rules (LLM judge, code, human).' - name: Experiments description: >- Experiments: offline runs of a pinned dataset snapshot through a target, scored by pinned evaluators. - name: Annotation queues description: >- Human review: queue traces, threads and experiment items, reserve one at a time, score them and feed corrections back into datasets. - name: Monitors description: >- Monitors: evaluators scoring a sampled share of a deployment's live traffic. - name: Observability description: Agent conversations, agents and their traffic. - name: Public runs description: Unauthenticated reads of runs their owners published to Explore. paths: /v1/deployments/{deployment_id}/adapters/{serve_name}/rollback: post: tags: - Deployments summary: Roll back an adapter description: >- Point the alias back at the previously served adapter model (previous_model_id) under a fresh version. Status returns to 'attaching' until every serving replica reports the rollback target loaded: usually immediate when the old weights are still resident, otherwise after a re-download. operationId: rollback parameters: - name: deployment_id in: path description: Deployment ID required: true schema: type: string - name: serve_name in: path description: The adapter's public alias required: true schema: type: string responses: '200': description: >- The alias now serves the previous model at version+1, status 'attaching' until the fleet converges content: application/json: schema: $ref: '#/components/schemas/AdapterResponse' '400': description: Not a multi_adapter deployment '404': description: Deployment or adapter not found, or not owned by the caller '409': description: >- No previous version to roll back to, the previous model no longer exists, or the adapter changed concurrently security: - bearerAuth: [] components: schemas: AdapterResponse: type: object description: A LoRA adapter attached to a multi_adapter deployment. required: - object - id - deployment_id - serve_name - model_id - version - internal_name - status - created_at - updated_at - replicas properties: object: type: string description: Always "deployment_adapter". id: type: string description: Attachment ID (ada_...). deployment_id: type: string description: The deployment (fleet) this adapter is attached to. serve_name: type: string description: Public routing alias the adapter serves under. model_id: type: string description: The custom model currently served under this alias. version: type: integer format: int32 description: Monotonic version counter; incremented on every flip/rollback. internal_name: type: string description: |- vLLM-facing name ("{serve_name}@v{version}-{tag}", tag derived from model_id): never reused for different weights. status: type: string description: |- attaching | active | detaching. 'active' means every serving replica reports the current version loaded. previous_model_id: type: - string - 'null' description: The previously served model, if any (the rollback target). pending_model_id: type: - string - 'null' description: 'Mid-flip only: the staged new version awaiting fleet convergence.' pending_internal_name: type: - string - 'null' description: 'Mid-flip only: the staged version''s internal name.' model_version_id: type: - string - 'null' description: >- The platform model version (model_versions.id) currently served under this alias, when the model belongs to a lineage. pending_model_version_id: type: - string - 'null' description: |- Mid-flip only: the platform model version staged by a `production` alias move; the fleet cuts over to it automatically on convergence. created_at: type: string format: date-time description: Attachment creation time. updated_at: type: string format: date-time description: Last status/version change time. replicas: type: array items: $ref: '#/components/schemas/AdapterReplicaState' description: Per-serving-replica load state for the CURRENT internal_name. AdapterReplicaState: type: object description: >- Per-replica load state for one internal_name, rolled up into the GET list. required: - replica_id - state properties: replica_id: type: string description: Serving replica ID. state: type: string description: |- downloading | loading | loaded | failed, or "missing" when the replica's last report doesn't mention this internal_name at all. error: type: - string - 'null' description: Replica-reported error detail, present when state='failed'. securitySchemes: bearerAuth: type: http scheme: bearer bearerFormat: API key description: API key with the `vk_` prefix. Create one from the dashboard. ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.