Traictory tracks 18 models from xAI. The earliest entry is Grok-1.5 from March 2024; the most recent is Grok 4.6, released in August 2026. 6 of them shipped in the last twelve months.
The strongest xAI model in our data is Grok-3 Mini, with an average benchmark score of 87.8% across 4 published results, including 84.0% on GPQA and 96.0% on AIME 2024. Across the 13 xAI models with benchmark data the average is 67.7%.
13 of the models are available through an API, with input pricing from $0.20 to $3 per million tokens; Grok-4.1 Fast Non-Reasoning is the cheapest. The largest context window is 2M tokens on Grok-4.1 Fast Non-Reasoning. 15 of the 18 models accept images or other non-text input. None of the listed models publish open weights.
xAI Models
xAI Model Comparisons
Other Companies
.svg?dpl=dpl_3EZdmE8NdzRwHesETrvSRCczw81T){kind=icon}
xAI model catalog updated in 2026. See model comparisons and best models for coding.