evals.report
BenchmarksLabsCompareRun guidesIn the wild
xAIGrok

Grok 4.6

xAI · Grok. Released Aug 12, 2026.

Grok 4.6 is a model from xAI in the Grok family, released Aug 12, 2026. evals.report tracks 5 reported Grok 4.6 benchmark scores across FrontierCode, AA-Omniscience: Knowledge and Hallucination Benchmark, ARC-AGI-1, ARC-AGI-2, DeepSWE — each shown with its benchmark, metric, source status, and date, and never combined into a single ranking.

5 results

Benchmark results 5

Compare this model
BenchmarkCategoryScoreMetricStatusDate
FrontierCodeCoding48.0%weighted score (Main)OfficialAug 12, 2026Details
AA-Omniscience: Knowledge and Hallucination BenchmarkReasoning30AA-Omniscience IndexOfficialAug 12, 2026Details
ARC-AGI-1Reasoning87%accuracyOfficialAug 12, 2026Details
ARC-AGI-2Reasoning67.08%accuracyOfficialAug 12, 2026Details
DeepSWECoding65.9%% resolvedVerifiedAug 12, 2026Details