Model profile
Claude Fable 5
Evidence summary
Claude Fable 5 has an estimated overall rank of #2; its 90% source-sensitivity interval is #2–#12. Published evidence spans 16 evals and 7 of 7 behavior components. Its strongest relative result is Enkrypt AI Safety Leaderboard (bias_attack_non_success_rate, #1 of 260); its weakest is CAIS Risk Index (textquests_harm, #44 of 48).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗claude-fable-5
- OpenRouter ↗anthropic/claude-fable-5
- System card ↗Family-level model document · Anthropic · first party
- Release source ↗direct research preview identity date
- www.anthropic.com ↗
Published eval results
Rank is within that sub-eval. Black marks the observed result; the grey dot marks the value implied by the global rank. Values stay on each source’s native scale.
| Eval / sub-eval | Rank | Value | Distribution | Better | Source |
|---|---|---|---|---|---|
| AA-Omnisciencehallucination_rate | #61 / 311 | 0.5485 | ↓ lower | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #24 / 105 | 0.49 | ↑ higher | Source ↗official | |
| CAIS Risk Indexagent_red_teaming | #1 / 43 | 39.4 | ↓ lower | Source ↗official | |
| CAIS Risk Indexbioweapons_assistance | #1 / 48 | 0 | ↓ lower | Source ↗official | |
| CAIS Risk Indexhle_overconfidence | #2 / 49 | 25.5 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmachiavelli | #2 / 45 | 80.2 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmask | #7 / 51 | 7.2 | ↓ lower | Source ↗official | |
| CAIS Risk Indexpolitical_manipulation | #8 / 32 | 39.8 | ↓ lower | Source ↗official | |
| CAIS Risk Indextextquests_harm | #44 / 48 | 23.7 | ↓ lower | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #1 / 260 | 100 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #1 / 260 | 100 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #1 / 260 | 100 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #1 / 258 | 100 | ↑ higher | Source ↗official | |
| Gray Swan indirect prompt injection (15 attempts)attack_success_probability_k15_pct | #3 / 13 | 2.8 | ↓ lower | Source ↗official | |
| HUMAINE Trust, Ethics and Safetytrust_ethics_safety_score | #8 / 54 | 28.64 | ↑ higher | Source ↗official | |
| MACHIAVELLIdeception_relative_random_pct | #2 / 50 | 80.2 | ↓ lower | Source ↗official | |
| Manager Coercion Benchcoercion_ladder_depth | #7 / 31 | 7.1 | ↓ lower | Source ↗official | |
| Manager Coercion Benchfabrication_rate | #1 / 13 | 0 | ↓ lower | Source ↗official | |
| SM-Benchadversarial | #49 / 73 | 79.51 | ↑ higher | Source ↗official | |
| SM-Benchambiguous_interpretation | #4 / 73 | 92.56 | ↑ higher | Source ↗official | |
| SM-Benchanti_hallucination | #12 / 73 | 98.43 | ↑ higher | Source ↗official | |
| SM-Bencheq_boundaries | #61 / 73 | 53.65 | ↑ higher | Source ↗official | |
| SM-Benchoverfit | #63 / 73 | 34.43 | ↑ higher | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #83 / 102 | 6.46 | ↑ higher | Source ↗official | |
| SpeciEvalland_animal_4ns | #35 / 102 | 4.4 | ↓ lower | Source ↗official | |
| SpeciEvalsea_animal_4ns | #53 / 102 | 4.78 | ↓ lower | Source ↗official | |
| SpeciEvalspeciesism | #38 / 102 | 1.85 | ↓ lower | Source ↗official | |
| TACbase_welfare_rate | #3 / 68 | 55.77 | ↑ higher | Source ↗official |
Values evaluations
Descriptive values and political-framing results are separate from safety/ethics ranks. Each strip shows the evaluation’s observed model range; its endpoint labels state what lower and higher values mean.