Model profile
Evidence summary
Safety. Muse Spark has an estimated Safety rank of #93; its 90% source-sensitivity interval is #22–#308. Its behavior-only rank is #103; company governance moves the combined estimate to #93. Published Safety evidence spans 5 eval lineages and 3 of 7 components. Its strongest relative result is Humanity's Last Exam RMS calibration error (Scale Labs) (calibrationError, #10 of 42); its weakest is Arena Factuality — Text Arena (factuality-only weighting) (factuality_bt_rating, #109 of 111).
Freedom. Muse Spark has an estimated Freedom rank of #515; its 90% source-sensitivity interval is #349–#572. Published Freedom evidence spans 1 eval lineages and 1 of 1 components. Its strongest relative result is FORTRESS (average_risk_score, #32 of 50); its weakest is FORTRESS (average_risk_score, #32 of 50).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗muse-spark
- Official model page ↗Exact model document · Meta · first party
Safety evals
Rank and direction are specific to the Safety portfolio. Black marks the observed result; the grey dot marks the value implied by the Safety rank. Values stay on each source’s native scale.
| Eval / sub-eval | Rank | Value | Distribution | Source |
|---|---|---|---|---|
| AA-Omnisciencehallucination_rate | #206 / 345 | ↓0.8419 | Source ↗official | |
| Arena Factuality — Text Arena (factuality-only weighting)factuality_bt_rating | #109 / 111 | ↑1384.0 | Source ↗official | |
| FORTRESSaverage_risk_score | #19 / 50 | ↓20.22 | Source ↗official | |
| Gray Swan indirect prompt injection (15 attempts)attack_success_probability_k15_pct | #6 / 13 | ↓16.5 | Source ↗official | |
| Humanity's Last Exam RMS calibration error (Scale Labs)calibrationError | #10 / 42 | ↓50 | Source ↗official |
Freedom evals
Rank and direction are specific to the Freedom portfolio. Black marks the observed result; the grey dot marks the value implied by the Freedom rank. Values stay on each source’s native scale.