v2

latestOpenAPI 3.1.02026-08-05267431678.1 KB

inference-history

Get Inference Detail

Get a single inference record by ID.

Args: inference_id: The inference record UUID auth: Authentication result with user_id

Returns: InferenceRecord with full details

Raises: HTTPException: 404 if inference not found or outside the caller's team

get/inferences/{inference_id}

Path parameters

inference_idstring required

Response

Successful Response

idstring required

Unique inference ID

user_idstring required

User who made the inference

model_idstring required

Model ID used for inference

model_namestring nullable

Human-readable model name

taskstring nullable

Task type (legacy; may be NULL for new inferences)

inputstring required

Input text

{"stackTrail":"components:schemas:InferenceRecord:properties:output:anyOf","oasType":"schema","type":"unknown","title":"Output","description":"Inference output (JSON)","nullable":true}

latency_msinteger nullable

End-to-end latency in milliseconds

ttft_msinteger nullable

Streaming time to first visible output chunk in milliseconds; null for non-streaming calls and streams with no visible payload.

tokensinteger nullable

Token count

input_tokensinteger nullable

Non-cached input/prompt tokens, sourced from the metered requests row. None when no billing row was recorded.

output_tokensinteger nullable

Output/completion tokens, sourced from the metered requests row.

cache_read_tokensinteger nullable

Input tokens served from the provider prompt cache (cache hit).

cache_write_tokensinteger nullable

Input tokens written into the provider prompt cache (cache creation). Zero for providers that bill writes as plain input.

sourcestring

Source of the request (api or ui)

statusstring

Inference status: success or failed

error_typestring nullable

Failure category when status is failed (validation, timeout, model_not_ready, model_not_found, model_not_supported, capacity_exhausted, internal)

error_messagestring nullable

Error detail when status is failed

created_atstring date-time required

When the inference was made

project_idstring nullable

Project ID the model belongs to

training_job_idstring nullable

Training job UUID that produced the model

providerstring nullable

Inference provider (aws, modal, fireworks, etc.)

base_modelstring nullable

HuggingFace base model ID

metadataobject nullable

Extensible metadata (e.g. LLM judge results)

human_verdictstring nullable

Human reviewer verdict (correct/incorrect)

{"stackTrail":"components:schemas:InferenceRecord:properties:human_corrected_output:anyOf","oasType":"schema","type":"unknown","title":"Human Corrected Output","description":"Expected output from human reviewer","nullable":true}

human_feedback_notesstring nullable

Optional reviewer notes

human_feedback_atstring date-time nullable

When human feedback was submitted

llmaj_verdict'pass' | 'fail' | 'uncertain' nullable

LLMAJ judge verdict ('pass', 'fail', or 'uncertain'); None until judged.

llmaj_scorenumber nullable

LLMAJ judge confidence score in [0.0, 1.0]; None until judged.

llmaj_judged_atstring date-time nullable

Timestamp when LLMAJ judgment was recorded; None until judged.

llmaj_reasoningstring nullable

LLMAJ judge reasoning/explanation; None until judged.