Model profile
H2O Danube3 4B Chat
Evidence summary
Safety. H2O Danube3 4B Chat does not meet the evidence gate for a Safety rank. Published Safety evidence spans 1 eval lineages and 3 of 7 components. Its strongest relative result is Enkrypt AI Safety Leaderboard (cbrn_attack_non_success_rate, #94 of 270); its weakest is Enkrypt AI Safety Leaderboard (bias_attack_non_success_rate, #254 of 270).
Freedom. H2O Danube3 4B Chat has an estimated Freedom rank of #230; its 90% source-sensitivity interval is #92–#504. Published Freedom evidence spans 1 eval lineages and 1 of 1 components. Its strongest relative result is Enkrypt AI Safety Leaderboard (harmful_attack_non_success_rate, #44 of 270); its weakest is Enkrypt AI Safety Leaderboard (cbrn_attack_non_success_rate, #173 of 270).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- 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 |
|---|---|---|---|---|
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #254 / 270 | ↑6.98 | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #94 / 270 | ↑89.67 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #224 / 270 | ↑43.89 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #192 / 268 | ↑94.09 | 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 | #173 / 270 | ↓89.67 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #44 / 270 | ↓43.89 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #77 / 268 | ↓94.09 | Source ↗official |