Grok 5

xai· grokhealthy

Grok 5: curated reference profile, Proprietary license

Grok 5: curated reference profile drawn from the model's published reports. Not a Supermagin measurement. Run the official eval in the Studio to earn a verified score.

xAI+11% net improvement

Curated reference profile · not a Supermagin measurement

Model card & documentation

Grok 5

Curated reference profile · Proprietary license · v2026.07.19 · 1.6s latency

These figures are drawn from the model's published reports and are NOT a Supermagin measurement. Run the official eval in the Studio to earn a verified, re-verifiable score.

Capabilities

agent: 91%
reasoning: 90%
coding: 89%
multimodal: 85%

Notes

Low hallucination rate on audited evals.
Reference evals: SWE-Bench Verified, GPQA Diamond, MMLU-Pro, MATH-500, ARC-AGI-2, BigCodeBench, τ-bench.

Pooled score

credibility 100% · consistency 0.96

Score trend across attested runs

31204 total samples

More attested runs are needed to show a trend.

Category scores

reasoning

90%

coding

89%

agent

91%

multimodal

85%

cybersecurity

84%

Progress by year

2023

72%

2024

82%

2025

89%

2026

92%

Success rate

Failure rate

Hallucination

Avg latency

1600ms

per task

Avg cost

$0.0150

per task

Calibration gap

+0.10

positive = overconfident

Top failure signals

hallucination ×686logic_flaw ×1298instruction_failure ×974

Pooled brain diagnostics

Output Entropy

healthy

Top-Token Confidence

healthy

Participation Ratio

healthy

Condition Number

healthy

Gradient Chain Health

healthy

Failure rate by task type

code

12%10921

8%9361

reasoning

10%6241

classification

9%4681

Curated reference profile, descriptive data, not a Supermagin measurement.