// xAI Grok · Text LLM
grok-2: VRAM & local deployment
Text LLM model published by xAI.
released 2025-08-22licence: Licence not statedmemory figures are estimates
01Specs
| Publisher | xAI · xai-org |
|---|---|
| Family / series | xAI Grok · grok-2 |
| Type | Text LLM |
| Parameters | not stated in official metadata |
| Context length | 128K tokens (131,072)max_position_embeddings in config.json |
| Architecture | 64 layers · GQA · MoE |
| Licence | · Licence not statedThe model card metadata does not state a licence; check the repository files. |
| Released | 2025-08-22 (Hugging Face repo created) |
02How much VRAM it needs (estimate)
Total = weights + 10% + 0.5GB overhead + KV cache for one request (FP16 KV). More concurrent users multiply the KV column.
No estimate: the official metadata does not give a parameter count. Check the model card for tested hardware.
Bits per weight: BF16 16, FP8 8.5, 4-bit 4.85. “Fits” = total ≤ 90% of memory; Mac GPU can use ~2/3 of unified memory (≤36GB) or ~3/4 (larger). For MoE, all experts must be in memory, so size by total parameters. Method →
03How to run it
Only runtimes the publisher or the runtime’s official library names for this model.
- SGLang — mentioned in the official model card
04Official repositories & quantised builds
1 official repo(s) from xai-org.
- grok-2 main
06Sources
Data checked 2025-11-05. Parameters, context and licence are copied from the official repository; memory figures are our estimates.