xAImodel systemcard
Grok 4.5
#9of 27 · general
500kcontext
Proprietarynon-commercial
2026-07-16released
2026-02-01knowledge cutoff
Standing
dot = this model · shaded = the field's middle 60% · track = full fieldGeneral
57.7#9/27
Math
54.6#10/27
Coding
48.0#12/27
Health
51.6#13/26
Engineering
56.2#8/17
Analyst
40.0#7/11
Legal
44.7#16/26
Instruct
49.8#12/27
Agents
65.5#6/27
Professional
48.5#13/27
Multimodal
33.6#13/19
Hallucination
30.9#13/27
Speed
24.0#19/22
Value
59.5#6/27
Measured
≈ imputed · amber dot = lab-self-reported among the sourcesGPQA Diamond93.1%#7
HLE42.7%#12
MMLU-Pro89.2%#7
LiveBench75.8%#9
SWE-bench V86.6%#6
LiveCodeBench87.3%#8
SciCode54.0%#8
CritPt15.4%#13
Terminal-Bench74.7%#11
APEX Agents34.2%#10
GDPval51.3%#11
Arena1468#12
AA Index55.8#8
Vals Index65.3#9
Epoch Capabilities Index154.0#14
Operations
$2 / $6$ per 1M tokens, in / out
$3blended
56tokens / sec
$0.12$ per solved task
1468arena elo · 9,998 votes