Model profile
Grok 4.20 Multi Agent
xAIdeveloper
2026-03-10release date
#38 / 312overall rank
4eval lineages
Evidence summary
Grok 4.20 Multi Agent has an estimated overall rank of #38; its 90% source-sensitivity interval is #12–#138. Its behavior-only rank is #25; company governance moves the combined estimate to #38. Published evidence spans 4 evals and 4 of 7 behavior components. Its strongest relative result is SM-Bench (eq_boundaries, #3 of 79); its weakest is SM-Bench (adversarial, #58 of 79).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- OpenRouter ↗x-ai/grok-4.20-multi-agent
- Official model documentation ↗Family-level model document · xai · first party
- Release source ↗direct research preview identity date
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 |
|---|---|---|---|---|
| BullshitBench v2clear_pushback_rate | #12 / 106 | ↑0.655 | Source ↗official | |
| Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate | #47 / 104 | ↓33 | Source ↗official | |
| DystopiaBenchbasaglia_score | #22 / 50 | ↓60.2 | Source ↗official | |
| DystopiaBenchbaudrillard_score | #29 / 50 | ↓61.77 | Source ↗official | |
| DystopiaBenchhuxley_score | #31 / 50 | ↓74 | Source ↗official | |
| DystopiaBenchlaguardia_score | #23 / 50 | ↓65.3 | Source ↗official | |
| DystopiaBenchorwell_score | #33 / 50 | ↓72.9 | Source ↗official | |
| DystopiaBenchpetrov_score | #32 / 50 | ↓74.4 | Source ↗official | |
| SM-Benchadversarial | #58 / 79 | ↑79.02 | Source ↗official | |
| SM-Benchambiguous_interpretation | #5 / 79 | ↑92.86 | Source ↗official | |
| SM-Benchanti_hallucination | #31 / 79 | ↑94.76 | Source ↗official | |
| SM-Bencheq_boundaries | #3 / 79 | ↑83.99 | Source ↗official | |
| SM-Benchoverfit | #18 / 79 | ↑84.7 | Source ↗official |