# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved. # SPDX-License-Identifier: Apache-2.0 openapi: 3.0.3 info: title: NVIDIA Personal AI Router API version: 0.1.0 description: > HTTP interfaces exposed by a local NVIDIA Personal AI Router (PAIR) proxy. An application sends a normal engine-compatible request to the local proxy; PAIR selects one eligible node and forwards the entire request to that node's engine, then streams or returns the response. PAIR routes each request to a single node. It does not pool GPU memory, combine GPUs, split a model, or migrate an in-flight request. Use the exact base URL copied from PAIR's **Endpoints** window. Do not assume default engine ports such as `11434` (Ollama) or `1234` (LM Studio). servers: - url: "{baseUrl}" description: > Local PAIR proxy base URL. Copy the exact value from PAIR's Endpoints window for the running engine you want to target. variables: baseUrl: default: http://127.0.0.1:11434 description: > Placeholder only. The real host and port are assigned by PAIR and shown in the Endpoints window; do not assume default engine ports. tags: - name: Ollama-compatible description: Ollama-compatible inference proxied to an eligible node. - name: OpenAI-compatible description: OpenAI-compatible inference proxied to an eligible node. - name: Node info description: Read-only node telemetry polled from discovered PAIR nodes. paths: /api/chat: post: tags: - Ollama-compatible operationId: ollamaChat summary: Ollama-compatible chat description: > Send an Ollama-compatible chat request to the local PAIR proxy. PAIR forwards the request to one eligible node running a compatible engine with the requested model available. With streaming enabled the upstream engine returns newline-delimited JSON chunks; the final chunk sets `done` to true. requestBody: required: true content: application/json: schema: $ref: "#/components/schemas/OllamaChatRequest" examples: default: value: model: llama3.1 stream: false messages: - role: user content: Explain what NVIDIA Personal AI Router does in one sentence. responses: "200": description: Chat response from the serving node. headers: X-PAIR-Served-By: description: Stable host UUID of the node that served the request. schema: type: string content: application/json: schema: $ref: "#/components/schemas/OllamaChatResponse" examples: default: value: model: llama3.1 created_at: "2026-06-03T07:45:00Z" message: role: assistant content: >- NVIDIA Personal AI Router routes each AI request to one eligible system on your local network to serve inference. done: true done_reason: stop total_duration: 4200000000 default: $ref: "#/components/responses/ErrorResponse" /v1/chat/completions: post: tags: - OpenAI-compatible operationId: createChatCompletion summary: OpenAI-compatible chat completion description: > Send an OpenAI-compatible chat completion request to the local PAIR proxy. PAIR forwards the entire request to a single eligible node and returns the engine's response. requestBody: required: true content: application/json: schema: $ref: "#/components/schemas/ChatCompletionRequest" examples: default: value: model: llama3.1 messages: - role: user content: Explain what NVIDIA Personal AI Router does in one sentence. responses: "200": description: Chat completion from the serving node. headers: X-PAIR-Served-By: description: Stable host UUID of the node that served the request. schema: type: string content: application/json: schema: $ref: "#/components/schemas/ChatCompletion" examples: default: value: id: chatcmpl-01J0D5P0KZESQHTR2HWK7S5W27 object: chat.completion created: 1780000000 model: llama3.1 choices: - index: 0 message: role: assistant content: >- NVIDIA Personal AI Router routes each AI request to one eligible system on your local network to serve inference. finish_reason: stop usage: prompt_tokens: 18 completion_tokens: 24 total_tokens: 42 default: $ref: "#/components/responses/ErrorResponse" /v1/node-info: get: tags: - Node info operationId: getNodeInfo summary: Get node info description: > Returns hardware inventory and measurements for a PAIR node. Service workers read public HTTP for unverified display previews and pinned mTLS for authenticated peer inventory. Desktop and TUI receive hardware through broker snapshots and updates. Verified host UUIDs correlate discovery, membership, workloads, telemetry, and routing; unverified previews use independent discovery record IDs. responses: "200": description: Node telemetry snapshot. content: application/json: schema: $ref: "#/components/schemas/NodeInfo" examples: default: value: uuid: 7c9e6679-7425-40de-944b-e07fc1f90ae7 hostname: workstation-01 addresses: - 192.168.1.24 - fe80::1c2d:3e4f:5a6b:7c8d servicePorts: proxy: 51100 broker: 51101 scanner: 51102 trust: trusted cluster: member: true clusterId: cluster_01J0D5TCGS8JQ2N8TZFRA0VN08 engines: - ollama models: - name: llama3.1 engine: ollama default: $ref: "#/components/responses/ErrorResponse" components: responses: ErrorResponse: description: Error response. content: application/json: schema: $ref: "#/components/schemas/Error" schemas: ChatRole: type: string enum: - system - user - assistant - tool ChatMessage: type: object required: - role - content properties: role: $ref: "#/components/schemas/ChatRole" content: type: string example: Explain what NVIDIA Personal AI Router does in one sentence. OllamaChatRequest: type: object required: - model - messages properties: model: type: string description: Model name that must be available on an eligible node. example: llama3.1 messages: type: array items: $ref: "#/components/schemas/ChatMessage" stream: type: boolean description: > When true (the Ollama default), the response is streamed as newline-delimited JSON chunks. default: true OllamaChatResponse: type: object required: - model - message - done properties: model: type: string example: llama3.1 created_at: type: string format: date-time message: $ref: "#/components/schemas/ChatMessage" done: type: boolean example: true done_reason: type: string example: stop total_duration: type: integer format: int64 description: Total request time in nanoseconds, as reported by the engine. example: 4200000000 ChatCompletionRequest: type: object required: - model - messages properties: model: type: string description: Model name that must be available on an eligible node. example: llama3.1 messages: type: array items: $ref: "#/components/schemas/ChatMessage" stream: type: boolean default: false temperature: type: number format: float minimum: 0 example: 0.7 max_tokens: type: integer minimum: 1 example: 512 ChatCompletion: type: object required: - id - object - created - model - choices properties: id: type: string example: chatcmpl-01J0D5P0KZESQHTR2HWK7S5W27 object: type: string example: chat.completion created: type: integer format: int64 description: Unix timestamp in seconds. example: 1780000000 model: type: string example: llama3.1 choices: type: array items: $ref: "#/components/schemas/ChatCompletionChoice" usage: $ref: "#/components/schemas/Usage" ChatCompletionChoice: type: object required: - index - message properties: index: type: integer example: 0 message: $ref: "#/components/schemas/ChatMessage" finish_reason: type: string enum: - stop - length - tool_calls - content_filter example: stop Usage: type: object properties: prompt_tokens: type: integer example: 18 completion_tokens: type: integer example: 24 total_tokens: type: integer example: 42 EngineKind: type: string enum: - ollama - lmstudio TrustState: type: string enum: - trusted - unpaired - pending - unknown NodeModel: type: object required: - name - engine properties: name: type: string example: llama3.1 engine: $ref: "#/components/schemas/EngineKind" ClusterState: type: object required: - member properties: member: type: boolean example: true clusterId: type: string nullable: true example: cluster_01J0D5TCGS8JQ2N8TZFRA0VN08 NodeInfo: type: object required: - uuid - hostname - trust - cluster properties: uuid: type: string description: Stable host UUID used to correlate discovery and routing. example: 7c9e6679-7425-40de-944b-e07fc1f90ae7 hostname: type: string example: workstation-01 addresses: type: array description: Reachable addresses advertised by the node. items: type: string example: - 192.168.1.24 servicePorts: type: object description: Service ports keyed by service name. additionalProperties: type: integer example: proxy: 51100 broker: 51101 trust: $ref: "#/components/schemas/TrustState" cluster: $ref: "#/components/schemas/ClusterState" engines: type: array items: $ref: "#/components/schemas/EngineKind" models: type: array description: Models reported as available on the node, where reported. items: $ref: "#/components/schemas/NodeModel" Error: type: object required: - code - message properties: code: type: string example: model_not_available message: type: string example: No eligible node has the requested model available. requestId: type: string example: req_01J0D5TCGS8JQ2N8TZFRA0VN08