Model profile
Claude Sonnet 5
Evidence summary
Claude Sonnet 5 has an estimated overall rank of #59; its 90% source-sensitivity interval is #2–#201. Its behavior-only rank is #76; company governance moves the combined estimate to #59. Published evidence spans 12 evals and 7 of 7 behavior components. Its strongest relative result is Enkrypt AI Safety Leaderboard (harmful_attack_non_success_rate, #1 of 241); its weakest is TAC (base_welfare_rate, #72 of 74).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗claude-sonnet-5
- OpenRouter ↗anthropic/claude-sonnet-5
- System card ↗Exact model document · Anthropic · first party
- Release source ↗curated outward evidence
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 | Source |
|---|---|---|---|---|
| AA-Omnisciencehallucination_rate | #38 / 327 | ↓0.3937 | Source ↗official | |
| Arena Factuality — Search Arena (factuality-only weighting)factuality_bt_rating | #14 / 30 | ↑1190.0 | Source ↗official | |
| Arena Factuality — Text Arena (factuality-only weighting)factuality_bt_rating | #22 / 112 | ↑1456.0 | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #5 / 106 | ↑0.79 | Source ↗official | |
| Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate | #2 / 104 | ↓2.243 | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #6 / 241 | ↑70.54 | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #43 / 241 | ↑92.5 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #1 / 241 | ↑100 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #9 / 239 | ↑99.82 | Source ↗official | |
| Gray Swan indirect prompt injection (15 attempts)attack_success_probability_k15_pct | #5 / 13 | ↓5.9 | Source ↗official | |
| Manager Coercion Benchcoercion_ladder_depth | #6 / 31 | ↓5.5 | Source ↗official | |
| Manager Coercion Benchfabrication_rate | #1 / 13 | ↓0 | Source ↗official | |
| Olam Social Poker — Social Lie Ratesocial_lie_rate_per_10000_turns | #3 / 19 | ↓2 | Source ↗official | |
| PHAREbias_resistance_diagnostic | #43 / 66 | ↑0.3988 | Source ↗official | |
| PHAREhallucination_resistance_diagnostic | #7 / 70 | ↑0.861 | Source ↗official | |
| PHAREharm_resistance_diagnostic | #4 / 70 | ↑0.9959 | Source ↗official | |
| PHAREjailbreak_resistance_diagnostic | #1 / 67 | ↑0.8684 | Source ↗official | |
| SM-Benchadversarial | #42 / 78 | ↑80.98 | Source ↗official | |
| SM-Benchambiguous_interpretation | #8 / 78 | ↑91.96 | Source ↗official | |
| SM-Benchanti_hallucination | #1 / 78 | ↑100 | Source ↗official | |
| SM-Bencheq_boundaries | #62 / 78 | ↑55.06 | Source ↗official | |
| SM-Benchoverfit | #19 / 78 | ↑83.06 | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #75 / 105 | ↑6.57 | Source ↗official | |
| SpeciEvalland_animal_4ns | #80 / 105 | ↓4.8 | Source ↗official | |
| SpeciEvalsea_animal_4ns | #87 / 105 | ↓5.05 | Source ↗official | |
| SpeciEvalspeciesism | #55 / 105 | ↓2.07 | Source ↗official | |
| TACbase_welfare_rate | #72 / 74 | ↑15.38 | 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.