Model profile
Hy3
Evidence summary
Safety. Hy3 has an estimated Safety rank of #88; its 90% source-sensitivity interval is #14–#256. Its behavior-only rank is #92; company governance moves the combined estimate to #88. Published Safety evidence spans 6 eval lineages and 5 of 7 components. Its strongest relative result is Enkrypt AI Safety Leaderboard (cbrn_attack_non_success_rate, #2 of 248); its weakest is TAC (base_welfare_rate, #79 of 87).
Freedom. Hy3 has an estimated Freedom rank of #629; its 90% source-sensitivity interval is #328–#641. Published Freedom evidence spans 2 eval lineages and 1 of 1 components. Its strongest relative result is Enkrypt AI Safety Leaderboard (toxicity_attack_non_success_rate, #136 of 246); its weakest is Enkrypt AI Safety Leaderboard (cbrn_attack_non_success_rate, #247 of 248).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗hy3
- Hugging Face ↗tencent/Hy3
- OpenRouter ↗tencent/hy3
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- Release source ↗repository created at proxy
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 | #140 / 345 | ↓0.7413 | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #23 / 248 | ↑43.93 | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #2 / 248 | ↑99.5 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #42 / 248 | ↑92.78 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #108 / 246 | ↑96.91 | Source ↗official | |
| Opposite-Narrator Sycophancysycophancy_rate_pct | #3 / 24 | ↓0.5 | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #30 / 123 | ↑6.967 | Source ↗self run | |
| SpeciEvalland_animal_4ns | #44 / 123 | ↓4.4 | Source ↗self run | |
| SpeciEvalsea_animal_4ns | #32 / 123 | ↓4.55 | Source ↗self run | |
| SpeciEvalspeciesism | #28 / 123 | ↓1.6 | Source ↗self run | |
| TACbase_welfare_rate | #79 / 87 | ↑16.67 | Source ↗self run | |
| WildClawBench Safety & Alignment (OpenClaw harness)safety_alignment_score_pct | #11 / 24 | ↑39 | 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.
| Eval / sub-eval | Rank | Value | Distribution | Source |
|---|---|---|---|---|
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #247 / 248 | ↓99.5 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #205 / 248 | ↓92.78 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #136 / 246 | ↓96.91 | Source ↗official | |
| SpeechMap model completioncomplete_pct | #160 / 181 | ↑34.2 | Source ↗official |