Model profile
Gemini 3.6 Flash
Evidence summary
Safety. Gemini 3.6 Flash has an estimated Safety rank of #37; its 90% source-sensitivity interval is #17–#130. Its behavior-only rank is #40; company governance moves the combined estimate to #37. Published Safety evidence spans 11 eval lineages and 6 of 7 components. Its strongest relative result is SpeciEval (land_animal_4ns, #2 of 131); its weakest is Enkrypt AI Safety Leaderboard (cbrn_attack_non_success_rate, #267 of 270).
Freedom. Gemini 3.6 Flash has an estimated Freedom rank of #30; its 90% source-sensitivity interval is #39–#304. Published Freedom evidence spans 3 eval lineages and 1 of 1 components. Its strongest relative result is Enkrypt AI Safety Leaderboard (cbrn_attack_non_success_rate, #4 of 270); its weakest is SM-Bench (adversarial, #86 of 92).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗gemini-3-6-flash
- OpenRouter ↗google/gemini-3.6-flash
- Official model documentation ↗Family-level model document · google · first party
- Release source ↗curated outward evidence
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 | #99 / 358 | ↓0.5563 | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #60 / 122 | ↑0.335 | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #130 / 270 | ↑16.28 | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #267 / 270 | ↑47.17 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #68 / 270 | ↑91.11 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #163 / 268 | ↑95.45 | Source ↗official | |
| Gray Swan indirect prompt injection (15 attempts)attack_success_probability_k15_pct | #10 / 13 | ↓37.3 | Source ↗official | |
| Olam Social Poker — Social Lie Ratesocial_lie_rate_per_10000_turns | #1 / 24 | ↓0 | Source ↗official | |
| Opposite-Narrator Sycophancysycophancy_rate_pct | #3 / 24 | ↓0.5 | Source ↗official | |
| Pander Scoreconversational_absolute_pander_score | #20 / 26 | ↓18.48 | Source ↗official | |
| Pander Scoreinstructional_absolute_pander_score | #19 / 26 | ↓66.01 | Source ↗official | |
| SM-Benchadversarial | #7 / 92 | ↑89.27 | Source ↗official | |
| SM-Benchambiguous_interpretation | #36 / 92 | ↑87.5 | Source ↗official | |
| SM-Benchanti_hallucination | #46 / 92 | ↑94.24 | Source ↗official | |
| SM-Bencheq_boundaries | #15 / 92 | ↑71.35 | Source ↗official | |
| SM-Benchoverfit | #5 / 92 | ↑95.08 | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #49 / 131 | ↑6.87 | Source ↗official | |
| SpeciEvalland_animal_4ns | #2 / 131 | ↓3.58 | Source ↗official | |
| SpeciEvalsea_animal_4ns | #46 / 131 | ↓4.65 | Source ↗official | |
| SpeciEvalspeciesism | #127 / 131 | ↓3.05 | Source ↗official | |
| Vals AI Cheating Auditbiomystery_bench_cheating_attempt_rate_pct | #8 / 9 | ↓7.778 | Source ↗official | |
| Vals Teen Conversation Safetyshare_flagged_pct | #6 / 8 | ↓31.94 | 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 | #4 / 270 | ↓47.17 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #199 / 270 | ↓91.11 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #101 / 268 | ↓95.45 | Source ↗official | |
| SM-Benchadversarial | #86 / 92 | ↓89.27 | Source ↗official | |
| SM-Bencheq_boundaries | #15 / 92 | ↑71.35 | Source ↗official | |
| SM-Benchoverfit | #5 / 92 | ↑95.08 | Source ↗official | |
| SpeechMap model completioncomplete_pct | #19 / 181 | ↑87.2 | Source ↗official |