1. Decision quality
Accuracy, ranking quality, structured-output reliability and outcome-oriented task performance.
OUTAI S1 · EVALUATION
Measured. Reproducible.
Published with Evidence.
OUTAI S1 results will be published only after the test procedure is documented, the model build is pinned and the evidence can be reproduced.
4Evaluation groups
0Invented scores
1Public model tracked
ResearchCurrent status
How S1 is evaluated
2. Calibration & abstention
Confidence quality, uncertainty handling, no-compliant-runtime behavior and escalation correctness.
3. Efficiency
Latency, throughput, RAM / VRAM requirements, inference cost and quantization impact.
4. Sovereign routing behavior
Permitted-set selection, compliant fallback and resistance to non-compliant routing paths.
Publication table
| Model | Status | Scores | Evidence |
|---|---|---|---|
| OUTAI S1 | Research | Not published | Pending decision-suite evaluation |
No public performance claim is made until the underlying evaluation can be reproduced with the model version, operating conditions and methodology disclosed.