> ## Documentation Index > Fetch the complete documentation index at: https://docs.bespokelabs.ai/llms.txt > Use this file to discover all available pages before exploring further. # System One > Send a state and questions about it to any Nimble model. The model field picks the model. Fact check models take the document as a string state and one noul question for each claim. Code search models take a query and a list of items in the state. The path /v1/nimble/systemone takes the same requests. ## OpenAPI ````yaml api-reference/openapi.yaml POST /v1/systemone openapi: 3.1.0 info: title: Bespoke Labs API version: '1.0' servers: - url: https://api.bespokelabs.ai security: - bearerAuth: [] paths: /v1/systemone: post: summary: System One description: >- Send a state and questions about it to any Nimble model. The model field picks the model. Fact check models take the document as a string state and one noul question for each claim. Code search models take a query and a list of items in the state. The path /v1/nimble/systemone takes the same requests. operationId: systemOne requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/SystemOneRequest' examples: questions: summary: Questions (nimble-latest) value: model: nimble-latest state: >- The customer was charged twice and asks for a refund. The service still works. questions: refund: type: noul instructions: Does the customer request a refund? department: type: choice instructions: Which department should handle this request? criteria: billing: Charges, payments and refunds technical: Software bugs and outages urgency: type: score instructions: How urgent is this request? criteria: - Routine billing request; service works - Some functionality unavailable - Complete service outage factcheck: summary: Fact check (nimble-factcheck) value: model: nimble-factcheck state: >- The board moved the vote to Tuesday because the CFO was travelling. questions: moved: type: noul instructions: The vote was moved to Tuesday. ceo: type: noul instructions: The CEO was travelling. codegrep: summary: Code search (nimble-codegrep) value: model: nimble-codegrep state: query: Uploads retry forever after a 413 response. items: - id: n0 path: src/upload/retry.py kind: file filePreview: text: |- def backoff(attempt): while True: retry() - id: n1 path: src/upload kind: directory childPreview: entries: - name: retry.py kind: file questions: q0: type: noul instructions: >- Does the file src/upload/retry.py contain code needed for this query? q1: type: noul instructions: >- Is the directory src/upload worth exploring for this query? responses: '200': description: The answers, keyed by question ID. headers: x-request-id: $ref: '#/components/headers/RequestId' x-typesafe-request-id: $ref: '#/components/headers/TypesafeRequestId' content: application/json: schema: $ref: '#/components/schemas/SystemOneResponse' examples: questions: summary: Questions (nimble-latest) value: model: nimble-latest answers: refund: type: noul noul: 0.998 department: type: choice choice: billing probabilities: billing: 1 technical: 0 confidence: 1 urgency: type: score score: 0 legend: '0': Routine billing request; service works '1': Some functionality unavailable '2': Complete service outage probabilities: '0': 1 '1': 0 '2': 0 confidence: 1 usage: input_tokens: 338 output_tokens: 4 factcheck: summary: Fact check (nimble-factcheck) value: model: nimble-factcheck answers: moved: type: noul noul: 0.99 ceo: type: noul noul: 0.002 usage: input_tokens: 25 output_tokens: 0 request_id: 6b57f82e-6905-48a1-bc44-fe0af138d8c5 codegrep: summary: Code search (nimble-codegrep) value: model: nimble-codegrep effort: medium answers: q0: type: noul noul: 0.99 q1: type: noul noul: 0.99 details: q0: raw: 0.85 overflow: false escalated: false scores: nimble-codegrep-4b: 0.85 q1: raw: 0.82 overflow: false escalated: false scores: nimble-codegrep-4b: 0.82 usage: input_tokens: 472 output_tokens: 2 request_id: 0789d4c7-399f-4e70-9721-9f21bab043bc escalation_skipped: false '401': description: The API key is missing or not valid. headers: x-request-id: $ref: '#/components/headers/RequestId' x-typesafe-request-id: $ref: '#/components/headers/TypesafeRequestId' content: application/json: schema: $ref: '#/components/schemas/Error' '402': description: Your organization does not have enough credit for the request. headers: x-request-id: $ref: '#/components/headers/RequestId' x-typesafe-request-id: $ref: '#/components/headers/TypesafeRequestId' content: application/json: schema: $ref: '#/components/schemas/Error' '413': description: The request body is larger than 1 MiB. headers: x-request-id: $ref: '#/components/headers/RequestId' x-typesafe-request-id: $ref: '#/components/headers/TypesafeRequestId' content: application/json: schema: $ref: '#/components/schemas/Error' '422': description: >- The request is not valid, or it is above a limit. The detail field says why. headers: x-request-id: $ref: '#/components/headers/RequestId' x-typesafe-request-id: $ref: '#/components/headers/TypesafeRequestId' content: application/json: schema: $ref: '#/components/schemas/Error' '429': description: >- Your organization already has 8 requests running. Retry after the number of seconds in the Retry-After header. headers: x-request-id: $ref: '#/components/headers/RequestId' x-typesafe-request-id: $ref: '#/components/headers/TypesafeRequestId' Retry-After: $ref: '#/components/headers/RetryAfter' content: application/json: schema: $ref: '#/components/schemas/Error' '502': description: A model returned an error. Retry with backoff. headers: x-request-id: $ref: '#/components/headers/RequestId' x-typesafe-request-id: $ref: '#/components/headers/TypesafeRequestId' content: application/json: schema: $ref: '#/components/schemas/Error' '503': description: >- A model is starting, or Nimble is not available for a moment. Retry after the number of seconds in the Retry-After header. You are not charged. headers: x-request-id: $ref: '#/components/headers/RequestId' x-typesafe-request-id: $ref: '#/components/headers/TypesafeRequestId' Retry-After: $ref: '#/components/headers/RetryAfter' content: application/json: schema: $ref: '#/components/schemas/Error' '504': description: The request took too long. headers: x-request-id: $ref: '#/components/headers/RequestId' x-typesafe-request-id: $ref: '#/components/headers/TypesafeRequestId' content: application/json: schema: $ref: '#/components/schemas/Error' '529': description: A general model is busy. Retry after about one second, with backoff. headers: x-request-id: $ref: '#/components/headers/RequestId' x-typesafe-request-id: $ref: '#/components/headers/TypesafeRequestId' content: application/json: schema: $ref: '#/components/schemas/Error' components: schemas: SystemOneRequest: description: >- The model field picks the model, and the model decides what the state and the questions must look like. oneOf: - $ref: '#/components/schemas/GeneralRequest' - $ref: '#/components/schemas/FactcheckRequest' - $ref: '#/components/schemas/CodegrepRequest' SystemOneResponse: type: object required: - model - answers - usage properties: model: type: string description: The model name from the request. answers: type: object description: Question IDs mapped to answers. additionalProperties: $ref: '#/components/schemas/Answer' usage: $ref: '#/components/schemas/Usage' request_id: type: string description: >- Fact check and code search only. The same ID as the x-request-id header. effort: type: string enum: - low - medium - high description: >- Code search only. low for nimble-codegrep-lite, medium for nimble-codegrep, and high for nimble-codegrep-max. details: type: object description: >- Code search only. Question IDs mapped to details about how each item was scored. additionalProperties: $ref: '#/components/schemas/CodegrepDetail' escalation_skipped: type: boolean description: Code search only. Always false in a 200 response. Error: type: object required: - detail - request_id properties: detail: description: Why the request failed. request_id: type: string description: The ID of this request, the same as the x-request-id header. GeneralRequest: type: object title: General models description: >- Typed questions about text or JSON, for nimble-latest, nimble-v3 or the previous model. required: - state - questions properties: model: type: string enum: - nimble-latest - nimble-v3 - bespokelabs/Bespoke-Nimble-9B-v3 - bespokelabs/Bespoke-Nimble-9B default: nimble-latest description: >- nimble-latest is the newest general model, which is nimble-v3 now. nimble-v3 and bespokelabs/Bespoke-Nimble-9B-v3 stay on this model, and bespokelabs/Bespoke-Nimble-9B is the previous model. The response repeats this name. state: description: >- The text or JSON to ask about. Nimble reads a JSON state as JSON text. oneOf: - type: string - type: object - type: array questions: type: object description: Your question IDs mapped to questions. The answers use the same IDs. minProperties: 1 maxProperties: 64 additionalProperties: $ref: '#/components/schemas/GeneralQuestion' FactcheckRequest: type: object title: Fact check description: Claims to check against a document. required: - model - state - questions properties: model: type: string enum: - nimble-factcheck-lite - nimble-factcheck - nimble-factcheck-max description: The fact check model. The response repeats this name. state: type: string maxLength: 400000 description: The document. questions: type: object description: Your claim IDs mapped to claims. The answers use the same IDs. minProperties: 1 maxProperties: 64 additionalProperties: $ref: '#/components/schemas/ClaimQuestion' split_claims: type: boolean default: true description: >- Score each sentence of a claim on its own and keep the lowest score. Set it to false to score each claim whole. CodegrepRequest: type: object title: Code search description: A query and the files and folders to score for it. required: - model - state - questions properties: model: type: string enum: - nimble-codegrep-lite - nimble-codegrep - nimble-codegrep-max description: The code search model. The response repeats this name. state: description: >- The query and the items to score, in at most 2,000,000 characters. Question q is about item n. A string state is accepted too, and Nimble uses it whole for every question, but the models were not trained on that. oneOf: - type: object required: - query - items properties: query: type: string description: What the coding agent is looking for. items: type: array description: The files and folders to score. items: type: object required: - id - path - kind properties: id: type: string description: n0, n1, and so on. Question q is about item n. path: type: string kind: type: string enum: - file - directory filePreview: type: object description: Part of a file's text. childPreview: type: object description: The entries of a folder. - type: string questions: type: object description: Your question IDs mapped to questions. The answers use the same IDs. minProperties: 1 maxProperties: 128 additionalProperties: $ref: '#/components/schemas/CodegrepQuestion' Answer: oneOf: - $ref: '#/components/schemas/NoulAnswer' - $ref: '#/components/schemas/ChoiceAnswer' - $ref: '#/components/schemas/ScoreAnswer' discriminator: propertyName: type mapping: noul: $ref: '#/components/schemas/NoulAnswer' choice: $ref: '#/components/schemas/ChoiceAnswer' score: $ref: '#/components/schemas/ScoreAnswer' Usage: type: object required: - input_tokens - output_tokens properties: input_tokens: type: integer description: The input tokens that you pay for. output_tokens: type: integer CodegrepDetail: type: object properties: raw: type: number description: The model's own probability before Nimble adjusts it. overflow: type: boolean description: >- True when the item's prompt was longer than 8,192 tokens. The answer is then 1.0, and the item is not charged. escalated: type: boolean description: Whether a larger model scored the item. scores: type: object description: Each model's score for the item. additionalProperties: type: number GeneralQuestion: oneOf: - $ref: '#/components/schemas/NoulQuestion' - $ref: '#/components/schemas/ChoiceQuestion' - $ref: '#/components/schemas/ScoreQuestion' discriminator: propertyName: type mapping: noul: $ref: '#/components/schemas/NoulQuestion' choice: $ref: '#/components/schemas/ChoiceQuestion' score: $ref: '#/components/schemas/ScoreQuestion' ClaimQuestion: type: object title: Claim description: >- One claim to check, as a noul question. Fact check refuses criteria and other fields. required: - type - instructions additionalProperties: false properties: type: type: string const: noul instructions: type: string minLength: 1 maxLength: 4000 description: The claim. CodegrepQuestion: oneOf: - $ref: '#/components/schemas/NoulQuestion' - $ref: '#/components/schemas/BooleanQuestion' discriminator: propertyName: type mapping: noul: $ref: '#/components/schemas/NoulQuestion' boolean: $ref: '#/components/schemas/BooleanQuestion' NoulAnswer: type: object title: Noul answer required: - type - noul properties: type: type: string const: noul noul: type: number minimum: 0 maximum: 1 description: >- The probability that the statement is true. For fact check, the probability that the document supports every part of the claim. For code search, the probability that the agent needs the item. ChoiceAnswer: type: object title: Choice answer required: - type - choice - probabilities - confidence properties: type: type: string const: choice choice: type: string description: The option with the highest probability. probabilities: type: object description: Every option's probability. They add up to 1. additionalProperties: type: number confidence: type: number description: >- 1 when one option has all the probability, 0 when every option is equally likely. It is not the chance that the answer is right. ScoreAnswer: type: object title: Score answer required: - type - score - legend - probabilities - confidence properties: type: type: string const: score score: type: number description: The expected level, counting from 0. It can fall between two levels. legend: type: object description: Level numbers mapped to their text. additionalProperties: type: string probabilities: type: object description: Every level's probability, keyed by level number. They add up to 1. additionalProperties: type: number confidence: type: number description: >- 1 when one level has all the probability, 0 when every level is equally likely. NoulQuestion: type: object title: Noul description: Asks whether a statement is true. required: - type properties: type: type: string const: noul instructions: type: string description: The statement or question. criteria: type: object description: What true and false mean. Optional. properties: 'true': type: string 'false': type: string ChoiceQuestion: type: object title: Choice description: Asks which option fits best. Only for the general models. required: - type - criteria properties: type: type: string const: choice instructions: type: string description: The question. criteria: type: object description: >- Option keys mapped to descriptions. If a description is null, Nimble reads the key instead. minProperties: 2 maxProperties: 255 additionalProperties: type: - string - 'null' ScoreQuestion: type: object title: Score description: Asks where the state falls on a scale. Only for the general models. required: - type - criteria properties: type: type: string const: score instructions: type: string description: The question. criteria: type: array description: The levels, from lowest to highest. minItems: 2 maxItems: 255 items: type: string BooleanQuestion: type: object title: Boolean description: >- Only for code search. Nimble answers it the same way as a noul question, with a noul answer. required: - type properties: type: type: string const: boolean instructions: type: string description: The question. headers: RequestId: description: >- The ID of this request. Send it to Bespoke Labs when you report a problem. schema: type: string TypesafeRequestId: description: The same ID as x-request-id. TypeSafe's SDK reads this header. schema: type: string RetryAfter: description: The number of seconds to wait before you retry. schema: type: integer securitySchemes: bearerAuth: type: http scheme: bearer description: Your Bespoke Labs API key. ````