Model profile
Claude Fable 5.1
Evidence summary
Safety. Claude Fable 5.1 has an estimated Safety rank of #4; its 90% source-sensitivity interval is #1–#202. Its behavior-only rank is #8; company governance moves the combined estimate to #4. Published Safety evidence spans 13 eval lineages and 7 of 7 components. Its strongest relative result is CAIS Risk Index (hle_overconfidence, #1 of 55); its weakest is Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibration (disordered_eating_claude_ai_harmless_rate_pct, #4 of 4).
Freedom. Claude Fable 5.1 has an estimated Freedom rank of #604; its 90% source-sensitivity interval is #325–#627. Published Freedom evidence spans 4 eval lineages and 1 of 1 components. Its strongest relative result is Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibration (child_claude_ai_overrefusal_rate_pct, #1 of 4); its weakest is Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibration (child_claude_ai_multiturn_appropriate_rate_pct, #4 of 4).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗05776db7-f5c0-40f7-b824-079160f8cfa8
- System card ↗Exact model document · Anthropic · first party
- artificialanalysis.ai ↗
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 | #106 / 338 | ↓0.6563 | Source ↗official | |
| CAIS Risk Indexagent_red_teaming | #3 / 49 | ↓34.2 | Source ↗official | |
| CAIS Risk Indexbioweapons_assistance | #2 / 54 | ↓0.5 | Source ↗official | |
| CAIS Risk Indexhle_overconfidence | #1 / 55 | ↓20.3 | Source ↗official | |
| CAIS Risk Indexmachiavelli | #10 / 51 | ↓82.4 | Source ↗official | |
| CAIS Risk Indexmask | #22 / 57 | ↓14.1 | Source ↗official | |
| CAIS Risk Indexpolitical_manipulation | #4 / 48 | ↓33.4 | Source ↗official | |
| CAIS Risk Indextextquests_harm | #52 / 54 | ↓27.1 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationchild_claude_ai_harmless_rate_pct | #3 / 4 | ↑99.98 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationchild_claude_ai_multiturn_appropriate_rate_pct | #1 / 4 | ↑100 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationchild_claude_ai_overrefusal_rate_pct | #1 / 4 | ↓0.17 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationdisordered_eating_claude_ai_harmless_rate_pct | #4 / 4 | ↑99.43 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationdisordered_eating_claude_ai_overrefusal_rate_pct | #1 / 4 | ↓0 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationelection_claude_ai_harmless_rate_pct | #1 / 4 | ↑100 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationelection_claude_ai_multiturn_appropriate_rate_pct | #3 / 4 | ↑88 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationelection_claude_ai_overrefusal_rate_pct | #1 / 4 | ↓0 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationmultiturn_biological_weapons_claude_ai_appropriate_rate_pct | #2 / 4 | ↑89 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationmultiturn_cyberattacks_claude_ai_appropriate_rate_pct | #2 / 4 | ↑99 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationmultiturn_deadly_weapons_claude_ai_appropriate_rate_pct | #3 / 4 | ↑89 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationmultiturn_hate_and_discrimination_claude_ai_appropriate_rate_pct | #4 / 4 | ↑95 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationmultiturn_influence_operations_claude_ai_appropriate_rate_pct | #1 / 4 | ↑76 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationmultiturn_romance_scams_claude_ai_appropriate_rate_pct | #3 / 4 | ↑94 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationmultiturn_tracking_and_surveillance_claude_ai_appropriate_rate_pct | #4 / 4 | ↑83 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationmultiturn_violent_extremism_claude_ai_appropriate_rate_pct | #2 / 4 | ↑93 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationoverall_benign_claude_ai_overrefusal_rate_pct | #1 / 4 | ↓0.34 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationoverall_harmful_claude_ai_harmless_rate_pct | #2 / 4 | ↑99.53 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationpolitical_claude_ai_refusal_rate_pct | #2 / 4 | ↓5 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationpolitical_evenhandedness_claude_ai_pct | #3 / 4 | ↑98.8 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationpolitical_opposing_perspectives_claude_ai_pct | #1 / 4 | ↑80.5 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationselfharm_claude_ai_harmless_rate_pct | #1 / 4 | ↑100 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationselfharm_claude_ai_multiturn_appropriate_rate_pct | #2 / 4 | ↑94 | Source ↗official | |
| Claude Fable 5.1 / Mythos 5.1 card — safety and refusal calibrationselfharm_claude_ai_overrefusal_rate_pct | #2 / 4 | ↓0.45 | Source ↗official | |
| Claude Fable 5.1 card — Gray Swan indirect prompt injection k=15attack_success_probability_k15_pct | #1 / 11 | ↓1 | Source ↗official | |
| FORTRESSaverage_risk_score | #5 / 50 | ↓13.72 | Source ↗official | |
| FORTRESSover_refusal_score | #47 / 49 | ↓16.91 | Source ↗official | |
| GPT 6 Astra system-card alignment evaluationsinternal_computer_use_safety_error_rate_pct | #2 / 5 | ↓9.5 | Source ↗official | |
| Humanity's Last Exam RMS calibration error (Scale Labs)calibrationError | #1 / 40 | ↓20 | Source ↗official | |
| SM-Benchadversarial | #24 / 84 | ↑85.37 | Source ↗official | |
| SM-Benchambiguous_interpretation | #5 / 84 | ↑92.86 | Source ↗official | |
| SM-Benchanti_hallucination | #29 / 84 | ↑96.34 | Source ↗official | |
| SM-Bencheq_boundaries | #31 / 84 | ↑67.7 | Source ↗official | |
| SM-Benchoverfit | #22 / 84 | ↑84.7 | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #108 / 122 | ↑6.37 | Source ↗official | |
| SpeciEvalland_animal_4ns | #53 / 122 | ↓4.47 | Source ↗official | |
| SpeciEvalsea_animal_4ns | #17 / 122 | ↓4.4 | Source ↗official | |
| SpeciEvalspeciesism | #61 / 122 | ↓2 | 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.