Model profile
Llama 3.2 90B Vision Instruct
Metadeveloper
2024-09-19release date
#205 / 309overall rank
4eval lineages
Evidence summary
Llama 3.2 90B Vision Instruct has an estimated overall rank of #205; its 90% source-sensitivity interval is #33–#285. Its behavior-only rank is #191; company governance moves the combined estimate to #205. Published evidence spans 4 evals and 4 of 7 behavior components. Its strongest relative result is Enkrypt AI Safety Leaderboard (bias_attack_non_success_rate, #13 of 241); its weakest is MASK (lying_probability_pct, #36 of 53).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Hugging Face ↗meta-llama/Llama-3.2-90B-Vision-Instruct
- NVIDIA ↗qc69jvmznzxy/llama-3.2-90b-vision-instruct
- OpenRouter ↗meta-llama/llama-3.2-90b-vision-instruct
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- Release source ↗repository created at proxy
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 | Source |
|---|---|---|---|---|
| Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate | #69 / 104 | ↓53.94 | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #13 / 241 | ↑52.2 | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #53 / 241 | ↑91.67 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #99 / 241 | ↑75 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #79 / 239 | ↑97.86 | Source ↗official | |
| MASKlying_probability_pct | #36 / 53 | ↓45.93 | Source ↗official | |
| SafeArenanormalized_safety_score | #3 / 5 | ↑34 | Source ↗official |