A WireMock simulation of Perplexity's REST API — covering OpenAI Chat Completions, OpenAI Responses and Anthropic Messages wire formats over Perplexity-hosted open-weight models, plus the model catalog and auth tokens.
A shared instance of this template runs on a public endpoint — call it right now from curl, an SDK, or an AI agent. Responses are fixed examples and behavior can't be changed; rate limits apply.
Jump to the endpoints ↓Run this template as a mock API in WireMock Cloud for a private instance you control — edit stubs, add stateful scenarios, use dynamic response templating, and inspect requests in the log.
Open in WireMock Cloud (free) →Live sandbox · no signup, no API key
A running WireMock Cloud instance of this template, callable right now from a terminal, a script, or an AI agent.
Endpoints
GET List Models
curl -s -X GET 'https://jg92lr.wiremockapi.cloud/orchestrator/v1/models'POST Create Chat Completion
curl -s -X POST 'https://jg92lr.wiremockapi.cloud/orchestrator/v1/chat/completions' \
-H 'Content-Type: application/json' \
-d '{}'GET Retrieve Model (creator/name path) - perplexity/deepseek-v4-pro-0813
curl -s -X GET 'https://jg92lr.wiremockapi.cloud/orchestrator/v1/models/perplexity/deepseek-v4-pro-0813'Spec catalog — Perplexity Orchestrator API (Chat 2.3.0 / Responses 2026-04-24 / Messages 0.5.0 / Auth 1.0.0) · 8 endpoints
| Method | Path | Summary |
|---|---|---|
| POST | /generate_auth_token | Generate Auth Token |
| POST | /orchestrator/v1/chat/completions | Create Chat Completion |
| POST | /orchestrator/v1/messages | Create Message |
| GET | /orchestrator/v1/models | List Models |
| GET | /orchestrator/v1/models/{creator}/{name} | Retrieve Model (creator/name path) |
| GET | /orchestrator/v1/models/{model} | Retrieve Model |
| POST | /orchestrator/v1/responses | Create Response |
| POST | /revoke_auth_token | Revoke Auth Token |
Unauthenticated WireMock Cloud demo sandbox for Perplexity — not an official Perplexity sandbox, and it returns simulated example data only.
The template simulates Perplexity's Orchestrator, which serves open-weight models through three wire formats at once. Its catalog is the four Perplexity-hosted models, each with per-million-token prices, and that catalog is enforced: a chat completion for perplexity/kimi-k3 answers a normal chat.completion object, while a request naming any other model gets an invalid_request_error pointing back at the models route. Router logic that validates model ids before sending can be tested against both outcomes.
Because the Orchestrator speaks the OpenAI wire format, the stock OpenAI SDK works once its base URL includes the /orchestrator/v1 prefix:
The Anthropic SDK connects the same way with its base URL set to the /orchestrator prefix (it appends /v1/messages itself), and agent frameworks that read OPENAI_BASE_URL can be redirected without code changes.
Fixed answers make multi-provider routing testable: switch formats, swap models and replay tool-call turns in CI with no preview access and no per-token cost.
from openai import OpenAI
# the sandbox URL, or your own mock's —
# the api_key is accepted but never validated
client = OpenAI(
base_url="https://jg92lr.wiremockapi.cloud/orchestrator/v1",
api_key="any-value",
)
reply = client.chat.completions.create(
model="perplexity/kimi-k3",
messages=[{"role": "user", "content": "Summarize WireMock in one line"}],
)
print(reply.choices[0].message.content)It quotes the last user turn, or the tool output when the last turn is a tool result. When tools are supplied, the reply is a tool call instead, to the tool named in tool_choice or else the first one, with arguments built from that tool's parameter schema.
Chat Completions and Responses return server-sent events when stream is true, word by word, with an optional usage chunk. Messages answers plain JSON even when streaming is requested, since the spec documents only a JSON response for that route.
No. Any key is accepted, and tokens from the generate route aren't checked on later calls. Keep real Perplexity keys away from the shared sandbox regardless.
The sandbox only simulates the unknown-model 400 and the unknown-model 404 on the catalog. Add 429 and 5xx stubs in your own WireMock Cloud copy to drive fallback logic between models.
Vendor names identify APIs represented by WireMock template sources. This page does not imply vendor endorsement, certification, partnership, or official integration status.