Model profile
Deepseek V4 Flash
Evidence summary
Deepseek V4 Flash has an estimated overall rank of #48; its 90% source-sensitivity interval is #39–#184. Its behavior-only rank is #41; company governance moves the combined estimate to #48. Published evidence spans 14 evals and 7 of 7 behavior components. Its strongest relative result is Manager Coercion Bench (coercion_ladder_depth, #1 of 31); its weakest is Inkling-Small model card — StrongREJECT (safety_rate, #10 of 10).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗deepseek-v4-flash
- Hugging Face ↗deepseek-ai/DeepSeek-V4-Flash
- NVIDIA ↗qc69jvmznzxy/deepseek-v4-flash
- OpenRouter ↗deepseek/deepseek-v4-flash
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- Release source ↗repository created at proxy
- huggingface.co ↗
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 | #252 / 311 | 0.8968 | ↓ lower | Source ↗official | |
| ANIMAscore | #2 / 19 | 0.7445 | ↑ higher | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #71 / 105 | 0.16 | ↑ higher | Source ↗official | |
| DystopiaBenchbasaglia_score | #23 / 50 | 60.9 | ↓ lower | Source ↗official | |
| DystopiaBenchbaudrillard_score | #27 / 50 | 58.1 | ↓ lower | Source ↗official | |
| DystopiaBenchhuxley_score | #27 / 50 | 72.8 | ↓ lower | Source ↗official | |
| DystopiaBenchlaguardia_score | #37 / 50 | 69.03 | ↓ lower | Source ↗official | |
| DystopiaBenchorwell_score | #28 / 50 | 69.83 | ↓ lower | Source ↗official | |
| DystopiaBenchpetrov_score | #30 / 50 | 74.3 | ↓ lower | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #148 / 260 | 14.73 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #246 / 260 | 70.33 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #106 / 260 | 76.67 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #204 / 258 | 91.64 | ↑ higher | Source ↗official | |
| HUMAINE Trust, Ethics and Safetytrust_ethics_safety_score | #10 / 54 | 28.57 | ↑ higher | Source ↗official | |
| Inkling-Small model card — FORTRESSbenign_answer_rate | #1 / 10 | 99.2 | ↑ higher | Source ↗official | |
| Inkling-Small model card — FORTRESSharmful_refusal_rate | #10 / 10 | 32 | ↑ higher | Source ↗official | |
| Inkling-Small model card — StrongREJECTsafety_rate | #10 / 10 | 97.4 | ↑ higher | Source ↗official | |
| Manager Coercion Benchcoercion_ladder_depth | #1 / 31 | 3.233 | ↓ lower | Source ↗self run | |
| MANTAAWMS | #4 / 7 | 0.417 | ↑ higher | Source ↗official | |
| MANTAAWVS | #3 / 7 | 0.508 | ↑ higher | Source ↗official | |
| PHAREbias_resistance_diagnostic | #59 / 66 | 0.323 | ↑ higher | Source ↗official | |
| PHAREhallucination_resistance_diagnostic | #13 / 70 | 0.8188 | ↑ higher | Source ↗official | |
| PHAREharm_resistance_diagnostic | #28 / 70 | 0.9507 | ↑ higher | Source ↗official | |
| PHAREjailbreak_resistance_diagnostic | #42 / 67 | 0.4549 | ↑ higher | Source ↗official | |
| SM-Benchadversarial | #36 / 73 | 81.95 | ↑ higher | Source ↗official | |
| SM-Benchambiguous_interpretation | #60 / 73 | 75.89 | ↑ higher | Source ↗official | |
| SM-Benchanti_hallucination | #30 / 73 | 94.24 | ↑ higher | Source ↗official | |
| SM-Bencheq_boundaries | #17 / 73 | 69.94 | ↑ higher | Source ↗official | |
| SM-Benchoverfit | #50 / 73 | 59.56 | ↑ higher | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #30 / 102 | 6.9 | ↑ higher | Source ↗official | |
| SpeciEvalland_animal_4ns | #67 / 102 | 4.7 | ↓ lower | Source ↗official | |
| SpeciEvalsea_animal_4ns | #49 / 102 | 4.75 | ↓ lower | Source ↗official | |
| SpeciEvalspeciesism | #53 / 102 | 2.05 | ↓ lower | Source ↗official | |
| TACbase_welfare_rate | #53 / 68 | 21.15 | ↑ higher | Source ↗official | |
| ToolPrivacyBenchprivate_mt_poi | #6 / 9 | 27.33 | ↓ lower | Source ↗official | |
| ToolPrivacyBenchpublic_mt_poi | #6 / 9 | 18.99 | ↓ lower | Source ↗official |
Values evaluations
Descriptive values and political-framing results are separate from safety/ethics ranks. Each strip shows the evaluation’s observed model range; its endpoint labels state what lower and higher values mean.