Model profile
Hy3
Evidence summary
Safety. Hy3 has an estimated Safety rank of #124; its 90% source-sensitivity interval is #20–#271. Its behavior-only rank is #120; company governance moves the combined estimate to #124. 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 270); its weakest is TAC (base_welfare_rate, #84 of 92).
Freedom. Hy3 has an estimated Freedom rank of #651; its 90% source-sensitivity interval is #357–#658. 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, #148 of 268); its weakest is Enkrypt AI Safety Leaderboard (cbrn_attack_non_success_rate, #269 of 270).
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 | #151 / 358 | ↓0.7413 | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #27 / 270 | ↑43.93 | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #2 / 270 | ↑99.5 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #60 / 270 | ↑92.78 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #117 / 268 | ↑96.91 | Source ↗official | |
| Opposite-Narrator Sycophancysycophancy_rate_pct | #3 / 24 | ↓0.5 | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #33 / 131 | ↑6.967 | Source ↗self run | |
| SpeciEvalland_animal_4ns | #47 / 131 | ↓4.4 | Source ↗self run | |
| SpeciEvalsea_animal_4ns | #35 / 131 | ↓4.55 | Source ↗self run | |
| SpeciEvalspeciesism | #30 / 131 | ↓1.6 | Source ↗self run | |
| TACbase_welfare_rate | #84 / 92 | ↑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 | #269 / 270 | ↓99.5 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #209 / 270 | ↓92.78 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #148 / 268 | ↓96.91 | Source ↗official | |
| SpeechMap model completioncomplete_pct | #160 / 181 | ↑34.2 | Source ↗official |