> ## 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. # Get a lineage's promotion gate > The gate config (dataset snapshot, pinned evaluator versions, thresholds, trials, error and valid-item limits, non-regression), the current production version's latest matching experiment (the non-regression baseline), and for each pinned llm_judge evaluator how often it agrees with human scores on the same targets (a warning below 80% over at least 10 targets; never a block). ## OpenAPI ````yaml /api-reference/openapi.json get /v1/models/{name}/gate 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/models/{name}/gate: get: tags: - Models summary: Get a lineage's promotion gate description: >- The gate config (dataset snapshot, pinned evaluator versions, thresholds, trials, error and valid-item limits, non-regression), the current production version's latest matching experiment (the non-regression baseline), and for each pinned llm_judge evaluator how often it agrees with human scores on the same targets (a warning below 80% over at least 10 targets; never a block). operationId: get_gate parameters: - name: name in: path description: Lineage name required: true schema: type: string responses: '200': description: The gate content: application/json: schema: $ref: '#/components/schemas/ModelGateResponse' '400': description: Evals disabled '404': description: No lineage with this name in the caller's workspace security: - bearerAuth: [] components: schemas: ModelGateResponse: type: object description: A lineage's promotion gate. required: - object - model - enabled - judge_human_agreement properties: object: type: string description: Always "model_gate". model: type: string description: Lineage name. enabled: type: boolean description: While true, moving `production` needs evidence that meets `config`. config: oneOf: - type: 'null' - $ref: '#/components/schemas/GateConfig' description: null until configured. Kept when the gate is disabled. updated_at: type: - string - 'null' format: date-time updated_by: type: - string - 'null' latest_production_experiment: oneOf: - type: 'null' - $ref: '#/components/schemas/ExperimentResponse' description: >- The current production version's latest completed run on the gate snapshot with the gate's evaluator pins: the non-regression baseline. judge_human_agreement: type: array items: $ref: '#/components/schemas/JudgeHumanAgreement' description: |- Per pinned llm_judge evaluator: agreement with human scores on the same targets. Informational; never blocks a promotion. GateConfig: type: object description: What promotion evidence must look like (`models.gate_config`). properties: dataset_snapshot_id: type: - string - 'null' description: The snapshot the evidence ran on (`@snap-N`). evaluators: type: array items: $ref: '#/components/schemas/GatePin' description: The exact evaluator versions the evidence was scored by. thresholds: type: object description: >- evaluator_id -> minimum pass rate in [0, 1] (all items in the denominator). Pinned evaluators with a pass rule and no entry need 1.0. additionalProperties: type: number format: double propertyNames: type: string trials: type: integer format: int32 description: The evidence ran at least this many trials per row. max_error_rate: type: number format: double description: Maximum errors / items per evaluator, in [0, 1]. min_valid_items: type: integer format: int64 description: Minimum ok scores per evaluator. non_regression: type: boolean description: |- Also require no evaluator to be worse than the current production version's run on the same snapshot and pins. regression_tolerance: type: number format: double description: |- How much worse (pass-rate points in [0, 1]; for a numeric evaluator without a pass rule, that fraction of its range) still counts as no regression. ExperimentResponse: type: object description: One experiment. required: - object - id - name - status - dataset_id - dataset_snapshot_id - target - evaluators - bindings - generation - trials - items_total - items_done - created_at properties: object: type: string description: Always "experiment". id: type: string description: '`exp_...`' name: type: string status: type: string description: queued | running | awaiting_local | completed | failed | canceled dataset_id: type: string dataset_snapshot_id: type: string description: The snapshot the run is pinned to (`@snap-N`). target: type: object description: The target as requested. target_deployment_id: type: - string - 'null' description: |- Resolved at create: the deployment called, the request `model` sent (an adapter's versioned internal name for a model version on a fleet), the model version, and the served model's identity. target_model: type: - string - 'null' target_model_version_id: type: - string - 'null' target_served_model: type: - string - 'null' evaluators: type: array items: $ref: '#/components/schemas/EvaluatorPin' description: The pinned evaluator versions (+ each judge's served model). bindings: type: object generation: $ref: '#/components/schemas/Generation' trials: type: integer format: int32 items_total: type: integer format: int32 description: Items = rows x trials; done counts finished items (ok or error). items_done: type: integer format: int32 aggregates: type: - object - 'null' description: |- Per evaluator at completion: n, ok, errors, skipped, pass_rate (all items in the denominator), mean, label_counts, critical_failures. passed: type: - boolean - 'null' description: |- Every evaluator with a pass rule passed every item with no errors; null until completed, or when no evaluator has a pass rule. baseline_experiment_id: type: - string - 'null' diff: type: - object - 'null' description: The diff against the baseline, computed at completion. error: type: - string - 'null' created_by: type: - string - 'null' created_at: type: string format: date-time started_at: type: - string - 'null' format: date-time finished_at: type: - string - 'null' format: date-time JudgeHumanAgreement: type: object description: How often an llm_judge agrees with human scores on the same targets. required: - evaluator_id - version - name - count properties: evaluator_id: type: string version: type: integer format: int32 description: The pinned version whose (non-human) scores are compared. name: type: string count: type: integer format: int64 description: >- Targets with both a non-human score of the pinned version and a human score of the same evaluator (any version), both with a verdict. agreement: type: - number - 'null' format: double description: The fraction of those targets whose `passed` agrees; null when none. warning: type: - string - 'null' description: Set when count >= 10 and agreement < 0.8. Never blocks a promotion. GatePin: type: object description: One evaluator version the gate pins. required: - evaluator_id - version properties: evaluator_id: type: string version: type: integer format: int32 EvaluatorPin: type: object description: One pinned evaluator version, stored in `experiments.evaluators`. required: - evaluator_id - version properties: evaluator_id: type: string version: type: integer format: int32 name: type: string scope: type: string description: single_turn | thread method: type: string description: llm_judge | code judge_model: type: - string - 'null' description: 'llm_judge: the judge deployment''s served model id at create.' Generation: type: object description: Sampling params for the target's calls. Omitted = the server default. properties: temperature: type: - number - 'null' format: double max_tokens: type: - integer - 'null' format: int32 additionalProperties: false 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.