Skip to main content
GET
Get Evaluation Results

Authorizations

Authorization
string
header
required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Path Parameters

evaluation_id
string
required

Response

Returns detailed evaluation metrics and scores for a completed evaluation. Identity, status, and timestamps are served by GET /{evaluation_id}.

The evaluation's scored results. Identity, status and timestamps are served by GET /{evaluation_id}.

evaluation_id
string
required

Unique identifier for the evaluation

Example:

"evaluation_01k4x9m2p7q3r7b1"

raw_responses_count
integer
required

Number of raw responses generated; read metric means relative to this size

Example:

100

metrics
Metrics · object | null

Metric names and scores. binary_correctness_mean is the fraction of responses whose key facts match the benchmark's expected response (0 to 1). answer_relevance_mean rates how relevant each response is to its instruction (0 to 1) and ignores correctness; read the pair together. The metric set evolves with our evaluation research.

method
string | null

Evaluation method: 'eval' or 'eval-compare'

Example:

"eval-compare"

baseline_model_id
string | null

ID of baseline model (for comparison)

Example:

"model_01k4x9m2p7q3r9d2"

base_model
Base Model · object | null

Base model results (comparison mode)

Example:

"qwen-v2p5-0p5b-instruct"

eval_model
Eval Model · object | null

Eval model results (comparison mode)

Example:
comparison
Comparison · object | null

Comparison results between models

Example: