Private AI
Grok 4.5 is xAI's frontier reasoning model for coding, knowledge work, and STEM. It supports text, image, and file inputs with tool calling and structured outputs. Reasoning is always active. Content policy rejections can still be charged: xAI may pass through a $0.05 moderation-failure fee, or a $0.055 usage-guidelines violation fee, depending on which rejection upstream returns.
Added Jul 8, 2026
Context Window
500.0K
Max Output
500.0K
Input Price (Auto)
$2.10/1M
Output Price (Auto)
$6.30/1M
Cache Read (Auto)
$0.53/1M
Capabilities
Performance metrics and benchmarks
Sourced from Artificial Analysis.
Intelligence Index
53.8
Choose explicit providers for this model. Auto routing remains available as the default option.
Loading provider options…
Coding Index
72.4
GPQA Diamond
Graduate-level scientific reasoning
93.1%
Better than 99% of models compared
HLE
Humanity's Last Exam
40.3%
Better than 97% of models compared
AA-LCR
Long context reasoning evaluation
67.7%
Better than 88% of models compared
SciCode
Python programming for scientific computing
54.1%
Better than 97% of models compared
Last updated Jul 22, 2026
Artificial Analysis