v5

latestOpenAPI 3.1.02026-07-31129337684.6 KB

LLM Configs

Test Llm Config

Run a one-shot test invocation against an LLM configuration and return the model's reply.

The call runs entirely server-side against the persisted config so the (write-only, redacted) api_key never leaves the server. An optional inline config override lets the UI test unsaved edits; a redacted override (api_key=None) reuses the stored secret, mirroring update_llm_config.

Provider/runtime failures are returned as HTTP 200 with success=False and a clean error message (rather than a 500), so the dashboard can render them inline.

post/workflows/v1/llm-configs/{llm_config_id}/test

Path parameters

llm_config_idstring required

Request body

system_promptstring required

Required, non-empty system prompt.

Response

Successful Response

successboolean required

responsestring nullable

Concatenated model output text on success.

errorstring nullable

Clean error message when success is False.

providerstring nullable

modelstring nullable

prompt_tokensinteger nullable

completion_tokensinteger nullable

total_tokensinteger nullable

latency_msinteger nullable

backchannel_responsestring nullable

Backchannel text emitted before the main response, if any.

backchannel_prompt_tokensinteger nullable

backchannel_completion_tokensinteger nullable

backchannel_total_tokensinteger nullable