Keywords AI

Keywords AI is an LLM observability and gateway platform. It exposes an OpenAI-compatible proxy (chat completions across 250+ models) plus REST APIs for request logging, prompt management, threads, evaluations, and traces - all under a single Bearer-authenticated API at https://api.keywordsai.co/api. (The company is rebranding to Respan; the keywordsai.co host and API remain active.)

6 APIs 0 Features

AILLMObservabilityGatewayMonitoring

APIs

Keywords AI LLM Proxy (Chat Completions)

OpenAI-compatible chat completions proxy that routes requests to 250+ LLMs through one unified endpoint, with streaming (SSE), tool calling, structured outputs, fallbacks, load ...

Keywords AI Logging API

Asynchronous request-logging endpoint for ingesting LLM call telemetry (model, prompt/completion messages, tokens, cost, latency, customer identifier, metadata) when calls are n...

Keywords AI Prompts API

Manage prompts and prompt versions outside of application code - create, list, retrieve, update, and delete prompts; create, commit, and deploy versions with templated messages,...

Keywords AI Threads API

Query multi-turn conversation threads grouped from logged LLM calls, with filtering and pagination for reconstructing end-user sessions.

Keywords AI Evaluations API

Create and run evaluators to score LLM outputs - manage evaluator definitions and execute them against logs and datasets to measure quality, correctness, and safety.

Keywords AI Traces API

List, retrieve, and delete distributed traces (OpenTelemetry-aligned spans) capturing agent and multi-step LLM workflows for observability and debugging.

Collections

Keywords AI API

OPEN

Pricing Plans

Keywordsai Plans Pricing

3 plans

PLANS

Rate Limits

Keywordsai Rate Limits

4 limits

RATE LIMITS

FinOps

Keywordsai Finops

FINOPS

Event Specifications

Keywords AI Chat Completions Streaming (HTTP + SSE)

AsyncAPI 2.6 description of Keywords AI's **chat completion streaming** surface on the OpenAI-compatible gateway. Keywords AI does not publish a WebSocket API. The only asynchro...

ASYNCAPI

Resources

👥

GitHubOrganization

Sources

opencollection: 1.0.0
info:
  name: Keywords AI API
  version: '1.0'
request:
  auth:
    type: bearer
    token: '{{bearerToken}}'
items:
- info:
    name: Gateway
    type: folder
  items:
  - info:
      name: Create a chat completion
      type: http
    http:
      method: POST
      url: https://api.keywordsai.co/api/chat/completions
      body:
        type: json
        data: "{\n  \"model\": \"gpt-4o-mini\",\n  \"messages\": [{\"role\": \"user\", \"content\": \"Reply with exactly ok.\"\
          }],\n  \"max_tokens\": 16,\n  \"temperature\": 0\n}"
    docs: OpenAI-compatible chat completion through the Keywords AI gateway. Set stream=true for Server-Sent Events.
- info:
    name: Logging
    type: folder
  items:
  - info:
      name: Create a request log
      type: http
    http:
      method: POST
      url: https://api.keywordsai.co/api/request-logs/
      body:
        type: json
        data: "{\n  \"model\": \"gpt-4o\",\n  \"prompt_messages\": [{\"role\": \"user\", \"content\": \"Hello\"}],\n  \"completion_message\"\
          : {\"role\": \"assistant\", \"content\": \"Hi there!\"},\n  \"customer_identifier\": \"user_123\"\n}"
    docs: 'Asynchronously ingest LLM call telemetry. Legacy alias: /request-logs/create/.'
- info:
    name: Prompts
    type: folder
  items:
  - info:
      name: Create a prompt
      type: http
    http:
      method: POST
      url: https://api.keywordsai.co/api/prompts/
      body:
        type: json
        data: "{\n  \"name\": \"customer_support\"\n}"
    docs: Create a prompt.
  - info:
      name: List prompts
      type: http
    http:
      method: GET
      url: https://api.keywordsai.co/api/prompts/
    docs: List prompts.
  - info:
      name: Retrieve a prompt
      type: http
    http:
      method: GET
      url: https://api.keywordsai.co/api/prompts/:prompt_id/
      params:
      - name: prompt_id
        value: ''
        type: path
        description: The prompt id or slug
    docs: Retrieve a prompt by id or slug.
  - info:
      name: Create a prompt version
      type: http
    http:
      method: POST
      url: https://api.keywordsai.co/api/prompts/:prompt_id/versions/
      params:
      - name: prompt_id
        value: ''
        type: path
        description: The prompt id or slug
      body:
        type: json
        data: "{\n  \"model\": \"gpt-4o\",\n  \"messages\": [{\"role\": \"system\", \"content\": \"You are a helpful {{role}}\
          \ assistant.\"}],\n  \"temperature\": 0.7\n}"
    docs: Create a new version of a prompt with templated messages and model config.
- info:
    name: Threads
    type: folder
  items:
  - info:
      name: List threads
      type: http
    http:
      method: POST
      url: https://api.keywordsai.co/api/log_threads/
      body:
        type: json
        data: "{\n  \"page\": 1,\n  \"page_size\": 50\n}"
    docs: Retrieve conversation threads matching filters, with pagination.
- info:
    name: Traces
    type: folder
  items:
  - info:
      name: List traces
      type: http
    http:
      method: POST
      url: https://api.keywordsai.co/api/traces/list/
      body:
        type: json
        data: "{\n  \"page\": 1,\n  \"page_size\": 50\n}"
    docs: List traces with filtering and pagination.
  - info:
      name: Retrieve a trace
      type: http
    http:
      method: GET
      url: https://api.keywordsai.co/api/traces/:trace_id/
      params:
      - name: trace_id
        value: ''
        type: path
    docs: Retrieve a trace and its spans.
- info:
    name: Evaluators
    type: folder
  items:
  - info:
      name: Create an evaluator
      type: http
    http:
      method: POST
      url: https://api.keywordsai.co/api/evaluators/
      body:
        type: json
        data: "{\n  \"name\": \"answer_relevance\",\n  \"type\": \"llm\"\n}"
    docs: Create an evaluator.
  - info:
      name: List evaluators
      type: http
    http:
      method: POST
      url: https://api.keywordsai.co/api/evaluators/list/
      body:
        type: json
        data: '{}'
    docs: List evaluators matching filters.
  - info:
      name: Run an evaluator
      type: http
    http:
      method: POST
      url: https://api.keywordsai.co/api/evaluators/:evaluator_id/run/
      params:
      - name: evaluator_id
        value: ''
        type: path
      body:
        type: json
        data: "{\n  \"input\": \"What is the capital of France?\",\n  \"output\": \"Paris\"\n}"
    docs: Run an evaluator against supplied input/output and return scores.
- info:
    name: Users
    type: folder
  items:
  - info:
      name: List users (customers)
      type: http
    http:
      method: POST
      url: https://api.keywordsai.co/api/users/list/
      body:
        type: json
        data: '{}'
    docs: List end-user (customer) records with analytics and budget fields.
bundled: true