Model profile
Grok 4.3
Evidence summary
Grok 4.3 has an estimated overall rank of #98; its 90% source-sensitivity interval is #53–#182. Its behavior-only rank is #92; company governance moves the combined estimate to #98. Published evidence spans 19 evals and 7 of 7 behavior components. Its strongest relative result is AA-Omniscience (hallucination_rate, #4 of 311); its weakest is Enkrypt AI Safety Leaderboard (cbrn_attack_non_success_rate, #255 of 260).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗grok-4-3
- OpenRouter ↗x-ai/grok-4.3
- Official model documentation ↗Family-level model document · xai · first party
- Release source ↗curated outward evidence
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 | #4 / 311 | 0.1603 | ↓ lower | Source ↗official | |
| BioSecBench-Refusalbalanced_refusal_score | #9 / 10 | 0.1092 | ↑ higher | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #25 / 105 | 0.48 | ↑ higher | Source ↗official | |
| CAIS Risk Indexagent_red_teaming | #32 / 43 | 90.1 | ↓ lower | Source ↗official | |
| CAIS Risk Indexbioweapons_assistance | #13 / 48 | 60.1 | ↓ lower | Source ↗official | |
| CAIS Risk Indexhle_overconfidence | #9 / 49 | 42.5 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmachiavelli | #16 / 45 | 84.7 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmask | #39 / 51 | 48.3 | ↓ lower | Source ↗official | |
| CAIS Risk Indexpolitical_manipulation | #21 / 32 | 51.6 | ↓ lower | Source ↗official | |
| CAIS Risk Indextextquests_harm | #2 / 48 | 3.9 | ↓ lower | Source ↗official | |
| Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate | #46 / 105 | 32.95 | ↓ lower | Source ↗official | |
| DystopiaBenchbasaglia_score | #30 / 50 | 65.37 | ↓ lower | Source ↗official | |
| DystopiaBenchbaudrillard_score | #29 / 50 | 61.77 | ↓ lower | Source ↗official | |
| DystopiaBenchhuxley_score | #25 / 50 | 71.63 | ↓ lower | Source ↗official | |
| DystopiaBenchlaguardia_score | #45 / 50 | 71.3 | ↓ lower | Source ↗official | |
| DystopiaBenchorwell_score | #30 / 50 | 72.07 | ↓ lower | Source ↗official | |
| DystopiaBenchpetrov_score | #23 / 50 | 72.07 | ↓ lower | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #181 / 260 | 12.92 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #255 / 260 | 65.17 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #35 / 260 | 95 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #207 / 258 | 91.45 | ↑ higher | Source ↗official | |
| MACHIAVELLIdeception_relative_random_pct | #16 / 50 | 84.7 | ↓ lower | Source ↗official | |
| Manager Coercion Benchcoercion_ladder_depth | #13 / 31 | 8.2 | ↓ lower | Source ↗official | |
| Manager Coercion Benchfabrication_rate | #12 / 13 | 67 | ↓ lower | Source ↗official | |
| MANTAAWMS | #6 / 7 | 0.371 | ↑ higher | Source ↗official | |
| MANTAAWVS | #6 / 7 | 0.352 | ↑ higher | Source ↗official | |
| PHAREbias_resistance_diagnostic | #29 / 66 | 0.47 | ↑ higher | Source ↗official | |
| PHAREhallucination_resistance_diagnostic | #23 / 70 | 0.7946 | ↑ higher | Source ↗official | |
| PHAREharm_resistance_diagnostic | #36 / 70 | 0.9355 | ↑ higher | Source ↗official | |
| PHAREjailbreak_resistance_diagnostic | #28 / 67 | 0.545 | ↑ higher | Source ↗official | |
| SM-Benchadversarial | #28 / 73 | 83.41 | ↑ higher | Source ↗official | |
| SM-Benchambiguous_interpretation | #32 / 73 | 86.01 | ↑ higher | Source ↗official | |
| SM-Benchanti_hallucination | #15 / 73 | 97.91 | ↑ higher | Source ↗official | |
| SM-Bencheq_boundaries | #13 / 73 | 70.79 | ↑ higher | Source ↗official | |
| SM-Benchoverfit | #24 / 73 | 80.87 | ↑ higher | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #25 / 102 | 6.93 | ↑ higher | Source ↗official | |
| SpeciEvalland_animal_4ns | #54 / 102 | 4.58 | ↓ lower | Source ↗official | |
| SpeciEvalsea_animal_4ns | #29 / 102 | 4.65 | ↓ lower | Source ↗official | |
| SpeciEvalspeciesism | #68 / 102 | 2.25 | ↓ lower | Source ↗official | |
| TukaBenchafri_jbb_cultural_asr | #1 / 6 | 10.6 | ↓ lower | Source ↗official | |
| TukaBenchafri_jbb_harm_asr | #2 / 6 | 8.2 | ↓ lower | Source ↗official | |
| TukaBenchafrijail_mono_asr | #2 / 6 | 17.7 | ↓ 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.