Test fixture
Arithmetic, symbolic steps, and structured problem solving.
The model receives the prompt (and optional system message). The run uses scorer rubric_json_metrics with the JSON configuration below. Pass/fail and partial credit are determined entirely by that scorer against the model output; no human grading.
Return JSON only with keys answer, evidence, constraints. A client sends a request using API version 2. The server supports only version 1 and version 3. There is no backwards compatibility note. Should the server accept or reject the request?
{
"metrics": {
"accuracy": {
"checks": [
{
"contains": [
"reject"
]
},
{
"contains": [
"not supported"
]
},
{
"contains": [
"version 2"
]
}
]
},
"evidence": {
"checks": [
{
"contains": [
"version 1"
]
},
{
"contains": [
"version 3"
]
},
{
"contains": [
"no backwards"
]
}
]
},
"constraint": {
"checks": [
{
"contains": [
"unsupported version"
]
},
{
"contains": [
"no compatibility"
]
},
{
"contains": [
"strict version"
]
}
]
}
}
}temperature
0
max_tokens
500
timeout (s)
120
type
scored
file
reason-constraint-api-version-compat.json