Model profile
SeaLLMs v3 7B Chat
Evidence summary
Safety. SeaLLMs v3 7B Chat has an estimated Safety rank of #291; its 90% source-sensitivity interval is #111–#328. Its behavior-only rank is #289; company governance moves the combined estimate to #291. Published Safety evidence spans 3 eval lineages and 4 of 7 components. Its strongest relative result is ThaiSafetyBench (safety_score, #3 of 18); its weakest is Enkrypt AI Safety Leaderboard (bias_attack_non_success_rate, #203 of 248).
Freedom. SeaLLMs v3 7B Chat has an estimated Freedom rank of #397; its 90% source-sensitivity interval is #89–#595. 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, #70 of 246); its weakest is ThaiSafetyBench (safety_score, #16 of 18).
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 | #203 / 248 | ↑11.11 | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #166 / 248 | ↑84.67 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #145 / 248 | ↑61.67 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #175 / 246 | ↑93.82 | Source ↗official | |
| IndoBias-Pairs — parity-aware culturally grounded biasparity_score | #20 / 26 | ↑84.38 | Source ↗official | |
| ThaiSafetyBenchsafety_score | #3 / 18 | ↑90.17 | 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 | #80 / 248 | ↓84.67 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #101 / 248 | ↓61.67 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #70 / 246 | ↓93.82 | Source ↗official | |
| ThaiSafetyBenchsafety_score | #16 / 18 | ↓90.17 | Source ↗official |