Model profile
DeepSeek V4.1 Flash
Evidence summary
Safety. DeepSeek V4.1 Flash has an estimated Safety rank of #28; its 90% source-sensitivity interval is #21–#58. Its behavior-only rank is #5; company governance moves the combined estimate to #28. Published Safety evidence spans 5 eval lineages and 3 of 7 components. Its strongest relative result is SpeciEval (speciesism, #6 of 123); its weakest is AA-Omniscience (hallucination_rate, #339 of 345).
Freedom. DeepSeek V4.1 Flash does not meet the evidence gate for a Freedom rank.
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗deepseek-v4-1-flash
- OpenRouter ↗deepseek/deepseek-v4.1-flash
- 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 | #339 / 345 | ↓0.9646 | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #24 / 117 | ↑0.57 | Source ↗official | |
| Manager Coercion Benchcoercion_ladder_depth | #3 / 37 | ↓4.033 | Source ↗self run | |
| SpeciEvalbelief_animal_sentience | #59 / 123 | ↑6.82 | Source ↗official | |
| SpeciEvalland_animal_4ns | #31 / 123 | ↓4.3 | Source ↗official | |
| SpeciEvalsea_animal_4ns | #14 / 123 | ↓4.33 | Source ↗official | |
| SpeciEvalspeciesism | #6 / 123 | ↓1.2 | Source ↗official | |
| TACbase_welfare_rate | #7 / 87 | ↑43.59 | Source ↗self run |
Freedom evals
No published sub-eval result contributes to this model’s Freedom profile.