REST API reference

Endpoint groups

• Routing33 endpoints

  Route a request to a target and return the decision with its trace; optionally build a plan or execute it. Report outcomes for the learners and read statistics and the audit trail.

  • get/api/v1/audit
  • get/api/v1/evaluations
  • post/api/v1/evaluations
  • get/api/v1/evaluations/datasets
  • post/api/v1/evaluations/datasets
  • get/api/v1/evaluations/datasets/{dataset_id}
  • delete/api/v1/evaluations/datasets/{dataset_id}
  • post/api/v1/evaluations/datasets/{dataset_id}/results
  • post/api/v1/evaluations/datasets/{dataset_id}/rows
  • post/api/v1/evaluations/datasets/{dataset_id}/run
  • get/api/v1/evaluations/gates
  • post/api/v1/evaluations/gates
  • delete/api/v1/evaluations/gates/{gate_id}
  • post/api/v1/evaluations/gates/{gate_id}/run
  • get/api/v1/evaluations/judge
  • post/api/v1/evaluations/judge/run
  • post/api/v1/evaluations/replay
  • get/api/v1/evaluations/sample
  • get/api/v1/evaluations/{evaluation_id}
  • delete/api/v1/evaluations/{evaluation_id}
  • post/api/v1/feedback
  • get/api/v1/handoffs
  • get/api/v1/handoffs/{handoff_id}
  • post/api/v1/handoffs/{handoff_id}/cancel
  • post/api/v1/handoffs/{handoff_id}/resolve
  • post/api/v1/route
  • get/api/v1/shadow
  • post/api/v1/shadow
  • delete/api/v1/shadow
  • post/api/v1/shadow/promote
  • get/api/v1/shadow/{candidate_id}
  • get/api/v1/stats
  • get/api/v1/targets
• OpenAI-compatible15 endpoints

  OpenAI-compatible chat completions and model listing. Point an OpenAI SDK at the platform base URL and use model: "auto"; the router chooses the model per request.

  • post/v1/audio/speech
  • post/v1/audio/transcriptions
  • post/v1/chat/completions
  • post/v1/embeddings
  • post/v1/feedback
  • post/v1/images/generations
  • post/v1/messages
  • post/v1/messages/count_tokens
  • get/v1/models
  • post/v1/moderations
  • post/v1/rerank
  • post/v1/responses
  • post/v1/route
  • get/v1beta/models
  • post/v1beta/models/{model}:{action}
• Public endpoints44 endpoints

  No API key required: deployment information, the model catalogue with live statistics, rankings and self-serve signup.

  • get/api/v1/blog
  • get/api/v1/blog/{slug}
  • get/api/v1/blog/{slug}/image
  • get/api/v1/catalogue
  • post/api/v1/estimate
  • get/api/v1/facts
  • get/api/v1/growth/indexnow-key
  • get/api/v1/growth/linkedin/callback
  • get/api/v1/info
  • post/api/v1/journey
  • post/api/v1/journey/answer
  • get/api/v1/journey/blocker
  • get/api/v1/journey/reasons
  • get/api/v1/llms
  • get/api/v1/llms/{vendor}/{model}
  • get/api/v1/maintenance
  • get/api/v1/models
  • get/api/v1/models/recommended
  • get/api/v1/models/{model_id}
  • post/api/v1/newsletter/confirm
  • get/api/v1/newsletter/issues/{issue_id}
  • get/api/v1/newsletter/preferences
  • put/api/v1/newsletter/preferences
  • delete/api/v1/newsletter/preferences
  • post/api/v1/newsletter/subscribe
  • post/api/v1/newsletter/unsubscribe
  • get/api/v1/rankings
  • get/api/v1/rankings/listings
  • get/api/v1/rankings/listings/kinds
  • get/api/v1/rankings/models
  • get/api/v1/rankings/models/{vendor}/{model}
  • get/api/v1/rankings/targets
  • get/api/v1/rankings/targets/domains
  • post/api/v1/sales/inquiries
  • get/api/v1/services
  • get/api/v1/showcase
  • post/api/v1/signup
  • get/api/v1/site/photos
  • get/api/v1/site/photos/{key}
  • get/api/v1/stats/public
  • get/api/v1/status
  • get/api/v1/status/board
  • get/api/v1/status/incidents
  • get/api/v1/status/incidents/{incident_id}
• Assistant4 endpoints

  The support assistant: answers product questions from the documentation through the platform's own models with cited sources, keeps the conversation and hands over a ticket draft when a person is needed; no API key required.

  • post/api/v1/assistant/chat
  • get/api/v1/assistant/conversations/{conversation_id}
  • get/api/v1/assistant/kb/search
  • get/api/v1/assistant/status
• Account, keys and tenants97 endpoints

  The calling workspace: profile, usage, savings, activity log, API keys and tenants.

  • get/api/v1/activity
  • delete/api/v1/activity/logs
  • get/api/v1/alerts
  • get/api/v1/announcements
  • post/api/v1/announcements/{announcement_id}/dismiss
  • get/api/v1/cache
  • delete/api/v1/cache
  • post/api/v1/email/events/acs
  • get/api/v1/email/preferences
  • put/api/v1/email/preferences
  • post/api/v1/email/unsubscribe
  • get/api/v1/events
  • get/api/v1/events/stream
  • get/api/v1/governance
  • get/api/v1/governance/report
  • get/api/v1/insights
  • post/api/v1/insights/ask
  • get/api/v1/keys
  • post/api/v1/keys
  • get/api/v1/keys/scopes
  • patch/api/v1/keys/{key_id}
  • delete/api/v1/keys/{key_id}
  • post/api/v1/keys/{key_id}/rotate
  • get/api/v1/learning
  • get/api/v1/legal
  • get/api/v1/me
  • get/api/v1/me/audit
  • post/api/v1/me/delete
  • get/api/v1/me/email-preferences
  • put/api/v1/me/email-preferences
  • get/api/v1/me/export
  • get/api/v1/me/legal
  • post/api/v1/me/legal/accept
  • put/api/v1/me/login-alerts
  • get/api/v1/notifications
  • get/api/v1/notifications/channels
  • post/api/v1/notifications/channels
  • patch/api/v1/notifications/channels/{channel_id}
  • delete/api/v1/notifications/channels/{channel_id}
  • post/api/v1/notifications/channels/{channel_id}/test
  • get/api/v1/notifications/deliveries
  • post/api/v1/notifications/read
  • get/api/v1/notifications/unread
  • get/api/v1/onboarding
  • get/api/v1/policy
  • put/api/v1/policy
  • delete/api/v1/policy
  • get/api/v1/prompts
  • post/api/v1/prompts
  • get/api/v1/prompts/{name}
  • delete/api/v1/prompts/{name}
  • post/api/v1/prompts/{name}/render
  • get/api/v1/prompts/{name}/stats
  • get/api/v1/prompts/{name}/{version}
  • delete/api/v1/prompts/{name}/{version}
  • put/api/v1/prompts/{name}/{version}/labels
  • get/api/v1/quota
  • get/api/v1/referrals
  • post/api/v1/referrals/invite
  • get/api/v1/referrals/lookup
  • get/api/v1/savings
  • get/api/v1/signals
  • get/api/v1/sla
  • get/api/v1/telemetry/series
  • get/api/v1/tenants
  • put/api/v1/tenants
  • delete/api/v1/tenants/{slug}
  • get/api/v1/trace/{request_id}
  • get/api/v1/usage
  • get/api/v1/usage/activity
  • post/api/v1/usage/activity
  • get/api/v1/usage/activity/summary
  • get/api/v1/usage/imports
  • post/api/v1/usage/imports
  • get/api/v1/usage/imports/samples
  • post/api/v1/usage/imports/upload
  • get/api/v1/usage/imports/{import_id}
  • delete/api/v1/usage/imports/{import_id}
  • get/api/v1/usage/observed
  • post/api/v1/usage/observed
  • get/api/v1/usage/opportunity
  • get/api/v1/usage/report
  • get/api/v1/workspace/log-export
  • post/api/v1/workspace/log-export
  • patch/api/v1/workspace/log-export/{export_id}
  • delete/api/v1/workspace/log-export/{export_id}
  • post/api/v1/workspace/log-export/{export_id}/run
  • get/api/v1/workspace/mcp-servers
  • post/api/v1/workspace/mcp-servers
  • get/api/v1/workspace/mcp-servers/{server_id}
  • patch/api/v1/workspace/mcp-servers/{server_id}
  • delete/api/v1/workspace/mcp-servers/{server_id}
  • post/api/v1/workspace/mcp-servers/{server_id}/sync
  • get/api/v1/workspace/mcp-servers/{server_id}/tools
  • get/api/v1/workspace/reports
  • put/api/v1/workspace/reports
  • post/api/v1/workspace/reports/send
• Conversations12 endpoints

  What the routed chat remembers for the caller: conversations the platform keeps for /v1/chat/completions (osr.conversation_id continues one across desktop, browser and playground; the platform prepends the history and a rolling summary) and the long-term memory - facts the person stated about themselves, listed, edited, disabled or deleted here; PUT /memory/settings turns it off.

  • get/api/v1/conversations
  • delete/api/v1/conversations
  • get/api/v1/conversations/{conversation_id}
  • patch/api/v1/conversations/{conversation_id}
  • delete/api/v1/conversations/{conversation_id}
  • post/api/v1/conversations/{conversation_id}/split
  • get/api/v1/memory
  • post/api/v1/memory
  • delete/api/v1/memory
  • put/api/v1/memory/settings
  • patch/api/v1/memory/{memory_id}
  • delete/api/v1/memory/{memory_id}
• Knowledge11 endpoints

  The workspace's knowledge graph: typed, sourced facts about the people who ask, the targets, the topics and the sources, learned from the routed chat and added by hand; the engineered context placed before every chat (preview it with POST /knowledge/context), suggestions from the workspace's trained link predictor, and exports to JSON-LD, N-Triples, GraphML and Cypher. PUT /knowledge/settings turns it off or changes the token budget.

  • get/api/v1/knowledge
  • delete/api/v1/knowledge
  • post/api/v1/knowledge/context
  • get/api/v1/knowledge/export
  • get/api/v1/knowledge/facts
  • post/api/v1/knowledge/facts
  • delete/api/v1/knowledge/facts
  • post/api/v1/knowledge/forget
  • put/api/v1/knowledge/settings
  • get/api/v1/knowledge/suggest
  • post/api/v1/knowledge/train
• Releases4 endpoints

  Release notes parsed from the changelog: every released version with its sections and highlights, the version this deployment runs, and what a signed-in person has not read yet (What's new).

  • get/api/v1/me/whats-new
  • post/api/v1/me/whats-new/seen
  • get/api/v1/releases
  • get/api/v1/releases/{version}
• Content27 endpoints

  Content publishing for the workspace's own website: the feeds it follows and the briefs it sends become posts written through the router as its own metered requests; posts are reviewed, published and delivered to its destinations (a signed webhook, WordPress, Ghost) with a retried delivery log, or pulled from GET /content/posts and the RSS at GET /content/feed.xml. The content_* MCP tools do the same for agents.

  • get/api/v1/content
  • get/api/v1/content/deliveries
  • post/api/v1/content/deliveries/{delivery_id}/retry
  • get/api/v1/content/destinations
  • post/api/v1/content/destinations
  • patch/api/v1/content/destinations/{destination_id}
  • delete/api/v1/content/destinations/{destination_id}
  • post/api/v1/content/destinations/{destination_id}/test
  • get/api/v1/content/feed.xml
  • post/api/v1/content/feed/token
  • get/api/v1/content/feeds
  • post/api/v1/content/feeds
  • post/api/v1/content/feeds/preview
  • patch/api/v1/content/feeds/{feed_id}
  • delete/api/v1/content/feeds/{feed_id}
  • post/api/v1/content/feeds/{feed_id}/fetch
  • get/api/v1/content/posts
  • post/api/v1/content/posts
  • post/api/v1/content/posts/write
  • get/api/v1/content/posts/{post_id}
  • patch/api/v1/content/posts/{post_id}
  • delete/api/v1/content/posts/{post_id}
  • post/api/v1/content/posts/{post_id}/publish
  • post/api/v1/content/posts/{post_id}/regenerate
  • get/api/v1/content/profile
  • put/api/v1/content/profile
  • post/api/v1/content/refresh
• Workspaces28 endpoints

  Organization workspaces: members, roles, invitations, teams (rosters, leads, tenant bindings), organization SSO (browser sessions only) and the SCIM provisioning token.

  • patch/api/v1/workspace
  • delete/api/v1/workspace
  • get/api/v1/workspace/audit
  • post/api/v1/workspace/convert
  • post/api/v1/workspace/delete
  • post/api/v1/workspace/invites
  • delete/api/v1/workspace/invites/{invite_id}
  • post/api/v1/workspace/invites/{invite_id}/resend
  • get/api/v1/workspace/members
  • patch/api/v1/workspace/members/{user_id}
  • delete/api/v1/workspace/members/{user_id}
  • get/api/v1/workspace/scim
  • post/api/v1/workspace/scim/token
  • delete/api/v1/workspace/scim/token
  • get/api/v1/workspace/sso
  • put/api/v1/workspace/sso
  • delete/api/v1/workspace/sso
  • get/api/v1/workspace/teams
  • post/api/v1/workspace/teams
  • get/api/v1/workspace/teams/{slug}
  • patch/api/v1/workspace/teams/{slug}
  • delete/api/v1/workspace/teams/{slug}
  • post/api/v1/workspace/teams/{slug}/members
  • delete/api/v1/workspace/teams/{slug}/members/{user_id}
  • post/api/v1/workspace/transfer
  • get/api/v1/workspaces
  • post/api/v1/workspaces
  • post/api/v1/workspaces/{account_id}/switch
• Scim6 endpoints

  SCIM 2.0 (RFC 7643 / 7644) for the organization's identity provider - Entra ID, Okta, OneLogin and the like - authenticated with the provisioning token an owner minted under /api/v1/workspace/scim. Users are provisioned into, updated in and removed from the organization; Groups are not provisioned.

  • get/scim/v2/Users
  • post/scim/v2/Users
  • get/scim/v2/Users/{user_id}
  • put/scim/v2/Users/{user_id}
  • patch/scim/v2/Users/{user_id}
  • delete/scim/v2/Users/{user_id}
• Sign-in and sessions30 endpoints

  Sign-in with GitHub, Google or an organization identity provider; browser sessions; and CLI sign-in with device codes (osr login): the CLI requests a code, the person approves it in the browser, the CLI receives a workspace API key.

  • post/api/v1/auth/device/approve
  • post/api/v1/auth/device/code
  • post/api/v1/auth/device/deny
  • post/api/v1/auth/device/token
  • get/api/v1/auth/device/{user_code}
  • get/api/v1/auth/discover
  • post/api/v1/auth/email
  • get/api/v1/auth/identities
  • delete/api/v1/auth/identities/{provider}
  • get/api/v1/auth/invites/{token}
  • post/api/v1/auth/invites/{token}/accept
  • post/api/v1/auth/logout
  • delete/api/v1/auth/mfa
  • post/api/v1/auth/mfa/enable
  • post/api/v1/auth/mfa/recovery-codes
  • post/api/v1/auth/mfa/setup
  • post/api/v1/auth/mfa/verify
  • post/api/v1/auth/password
  • delete/api/v1/auth/password
  • post/api/v1/auth/password/forgot
  • post/api/v1/auth/password/login
  • post/api/v1/auth/password/reset
  • get/api/v1/auth/providers
  • get/api/v1/auth/sessions
  • delete/api/v1/auth/sessions/{session_id}
  • post/api/v1/auth/verify
  • post/api/v1/auth/verify/send
  • post/api/v1/auth/{provider}/callback
  • post/api/v1/auth/{provider}/credential
  • get/api/v1/auth/{provider}/start
• Billing16 endpoints

  Plan upgrades through Stripe Checkout and the webhook that confirms them.

  • post/api/v1/billing/cancel
  • post/api/v1/billing/checkout
  • get/api/v1/billing/coupons
  • post/api/v1/billing/coupons/redeem
  • post/api/v1/billing/coupons/validate
  • post/api/v1/billing/credits/checkout
  • get/api/v1/billing/invoices
  • post/api/v1/billing/portal
  • get/api/v1/billing/statement
  • get/api/v1/billing/subscription
  • get/api/v1/billing/usage
  • get/api/v1/billing/wallet
  • put/api/v1/billing/wallet/auto-topup
  • delete/api/v1/billing/wallet/auto-topup
  • get/api/v1/billing/wallet/ledger
  • post/api/v1/billing/webhook
• Support11 endpoints

  Help: support tickets (/support/tickets - raise a concern, get a reference such as OSR-00042, follow the conversation; every answer and status change is e-mailed to the requester), product feedback (/product-feedback - bugs, ideas and improvements sent from the platform's pages with screenshots and browser diagnostics, reference FB-00012, every status change e-mailed) and the assistant.

  • get/api/v1/product-feedback
  • post/api/v1/product-feedback
  • get/api/v1/product-feedback/{key}
  • get/api/v1/product-feedback/{key}/attachments/{attachment_id}
  • post/api/v1/product-feedback/{key}/comments
  • get/api/v1/support/tickets
  • post/api/v1/support/tickets
  • get/api/v1/support/tickets/{key}
  • post/api/v1/support/tickets/{key}/close
  • post/api/v1/support/tickets/{key}/messages
  • get/api/v1/support/tickets/{key}/stream
• MCP server5 endpoints

  Model Context Protocol server for IDEs and agents (VS Code, Cursor, Claude, Windsurf): POST /mcp speaks JSON-RPC with tools to route, quote, recommend, explain and ask; GET /mcp describes it. Authenticate like the REST API.

  • get/.well-known/oauth-protected-resource
  • get/.well-known/oauth-protected-resource/mcp
  • get/mcp
  • post/mcp
  • delete/mcp
• Marketplace17 endpoints

  Browse, publish, install and rate agents, skills, personas, prompts, templates and tools. Each listing has a validated manifest the router can load; GET /registry/{slug}/manifest is what registry:// references resolve to.

  • get/api/v1/me/installs
  • get/api/v1/me/listings
  • get/api/v1/registry
  • post/api/v1/registry
  • get/api/v1/registry/kinds
  • get/api/v1/registry/sitemap
  • get/api/v1/registry/{slug}
  • patch/api/v1/registry/{slug}
  • post/api/v1/registry/{slug}/archive
  • post/api/v1/registry/{slug}/install
  • get/api/v1/registry/{slug}/manifest
  • get/api/v1/registry/{slug}/reviews
  • put/api/v1/registry/{slug}/reviews
  • delete/api/v1/registry/{slug}/reviews
  • post/api/v1/registry/{slug}/submit
  • post/api/v1/registry/{slug}/unarchive
  • post/api/v1/registry/{slug}/versions
• Model providers1 endpoint

  Model providers: GET /providers is the public, OpenRouter-style view of what this deployment executes on (provider, upstream model, prices, health per executable target - never credentials); the /admin/providers endpoints add, check and map OpenAI-compatible endpoints at run time, and every API replica picks the change up without a restart.

  • get/api/v1/providers
• Offers2 endpoints

  Models by the token: an offer is one catalogue model served by one provider endpoint (vendor/model@provider), synthesised into the routing catalogue and sold at the customer price the pricing engine computes. GET /offers is the public storefront (list price next to the price the caller pays, the endpoints with health and traffic); /admin/offers and /admin/providers/{provider_id}/offers list, pause and remove offers. POST /v1/chat/completions accepts vendor/model (the router picks the endpoint) and vendor/model@provider (one pinned); GET /v1/models lists offers in OpenRouter's shape.

  • get/api/v1/offers
  • get/api/v1/offers/{vendor}/{model}
• Workspace-providers8 endpoints

  Bring your own model providers (feature byok, on every plan of the default catalogue): a workspace connects its own Azure OpenAI, Amazon Bedrock, Google Vertex AI / Gemini, OpenAI, Anthropic or any OpenAI-compatible endpoint with its own key (stored encrypted, never returned) and maps catalogue targets to models on it. Its /route (execute=true), /v1/chat/completions and MCP ask calls then run on that endpoint; unmapped targets keep the platform's providers. Reading needs membership, writing the admin role.

  • get/api/v1/workspace/providers
  • post/api/v1/workspace/providers
  • put/api/v1/workspace/providers/models/{target_id}
  • delete/api/v1/workspace/providers/models/{target_id}
  • get/api/v1/workspace/providers/{provider_id}
  • patch/api/v1/workspace/providers/{provider_id}
  • delete/api/v1/workspace/providers/{provider_id}
  • post/api/v1/workspace/providers/{provider_id}/check