Model profile
Gemma 4 E2B
2eval lineages
Evidence summary
Published evidence spans 2 evals and 4 of 7 behavior components. Its strongest relative result is AA-Omniscience (hallucination_rate, #21 of 311); its weakest is Enkrypt AI Safety Leaderboard (bias_attack_non_success_rate, #257 of 260).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗gemma-4-e2b
- Official model documentation ↗Family-level model document · google · first party
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 | #21 / 311 | 0.3308 | ↓ lower | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #257 / 260 | 1.55 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #249 / 260 | 68 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #69 / 260 | 86.67 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #128 / 258 | 96.36 | ↑ higher | Source ↗official |