How much VRAM does Grok 2.5 268B Chat need?
Grok 2.5 268B Chat needs about 159.6 GB of VRAM at Q4_K_M, or 279.2 GB at Q8_0 and 524.9 GB at FP16. No single GPU in our set holds it even at Q4_K_M; it needs multi-GPU or offload. Weights plus runtime overhead; with only 8K of context the cache stays small.
Weights-plus-runtime footprint across the six most common quantizations. KV cache is context-dependent and comes on top. The 8K window is short, so weights dominate the budget; size it precisely in the GPU Memory Calculator.
Can your GPU run Grok 2.5 268B Chat?
Pick your card — the answer below is computed by the same engine that produces every figure on this page.
H200 141GB cannot hold Grok 2.5 268B Chat at Q4_K_M: it needs 159.6 GB against 133.9 GB usable; a heavier quantization, a larger card, or splitting across GPUs are the ways forward; as a mixture-of-experts model all 268B parameters must be resident even though only a fraction ofB are active per token.
Weights plus runtime overhead. KV cache is not included — it depends on context length and this model’s published architecture.
| Weights (Q4_K_M) | 151.3 GB |
| Runtime overhead | 8.3 GB |
| Estimated total | 159.6 GB |
| Usable VRAM (95% of 141 GB) | 133.9 GB |
Across every GPU we track
- RTX 3060 12GBNot enough-148.2 GB
- RTX 4060 Ti 16GBNot enough-144.4 GB
- RTX 4070 Ti SUPERNot enough-144.4 GB
- RTX 3090Not enough-136.8 GB
- RTX 4090Not enough-136.8 GB
- RTX 5090Not enough-129.2 GB
- A100 40GBNot enough-121.6 GB
- RTX A6000Not enough-114.0 GB
- L40SNot enough-114.0 GB
- A100 80GBNot enough-83.6 GB
- H100 80GBNot enough-83.6 GB
- H200 141GBNot enough-25.7 GB
Estimates, not guarantees: real usage moves with runtime, driver, batch size and context. Size a specific context window in the GPU Memory Calculator.
How was this number calculated? Every figure here comes from Bitpute’s documented calculation methodology — parameters, precision, quantization, runtime overhead and usable VRAM, each formula written out in full.
Memory by quantization
| Quant | Weights | + runtime overhead |
|---|---|---|
| FP16 | 499.2 GB | 524.9 GB |
| Q8_0 | 265.2 GB | 279.2 GB |
| Q6_K | 204.7 GB | 215.7 GB |
| Q5_K_M | 177.5 GB | 187.2 GB |
| Q4_K_M | 151.3 GB | 159.6 GB |
| Q4_0 | 140.4 GB | 148.2 GB |
Overhead = 0.75 GB + 5% of weights (CUDA context, buffers). Add KV cache on top. With only 8K of context, weights dominate the budget.
Which quantization should you actually run?
Grok 2.5 268B Chat is a multi-GPU or datacenter model: even Q4_K_M needs about 159.6 GB, beyond any single card. Plan for several 80 GB GPUs or a hosted endpoint; FP16 (524.9 GB) is server-only.
Single-GPU fit at Q4_K_M (159.6 GB + KV)
RTX 3060 12GBRTX 4060 Ti 16GBRTX 4070 Ti SUPERRTX 3090RTX 4090RTX 5090A100 40GBRTX A6000L40SA100 80GBH100 80GBH200 141GB
Common questions
How much VRAM does Grok 2.5 268B Chat need?
Grok 2.5 268B Chat needs about 159.6 GB of VRAM at Q4_K_M, 279.2 GB at Q8_0, or 524.9 GB at FP16. That is weights plus about 0.75 GB of runtime overhead and 5% of weight size; KV cache is additional and depends on context length.
What GPU can run Grok 2.5 268B Chat?
No single GPU in our set fits it at Q4_K_M; it needs multiple GPUs or CPU offload. At FP16 no single GPU in our set is sufficient.
Can Grok 2.5 268B Chat run on 24 GB?
Not at Q4_K_M: it needs about 159.6 GB, more than the ~22.8 GB usable on a 24 GB card. Use a larger card, multiple GPUs, or offload.