Question 1

How much VRAM does MiMo-V2.5 need?

Accepted Answer

MiMo-V2.5 (310.8B) requires approximately 705.5821380615234 GB of VRAM at full precision (FP16), 352.8 GB at 8-bit quantization (Q8), and 176.4 GB at 4-bit quantization (Q4). The actual VRAM usage may be slightly higher due to KV cache and runtime overhead.

Question 2

What GPUs can run MiMo-V2.5?

Accepted Answer

We currently have no compatible GPUs indexed for MiMo-V2.5. This may be because the model's VRAM requirements exceed available GPUs in our database, or VRAM data hasn't been computed yet.

Question 3

What is the cheapest GPU to run MiMo-V2.5?

Accepted Answer

We don't currently have cloud pricing data for GPUs compatible with MiMo-V2.5. Cloud pricing data is updated regularly — check back soon.

Question 4

Can MiMo-V2.5 run on consumer hardware?

Accepted Answer

MiMo-V2.5 requires approximately 705.5821380615234 GB of VRAM at FP16, and even with 4-bit quantization (Q4) it needs about 176.4 GB. This exceeds the 24 GB available on the highest-end consumer GPU (RTX 4090), so MiMo-V2.5 requires datacenter or cloud GPUs.

Question 5

What quantization options are available for MiMo-V2.5?

Accepted Answer

MiMo-V2.5 supports the following quantization levels: FP16 (full precision): ~705.5821380615234 GB, Q8 (8-bit): ~352.8 GB, Q6 (6-bit): ~264.6 GB, Q4 (4-bit): ~176.4 GB. Lower quantization reduces VRAM usage at the cost of some quality loss. Q8 offers a good balance between quality and memory savings, while Q4 provides maximum compression. As a Mixture of Experts (MoE) model with 256 experts (8 active), MiMo-V2.5 has a larger total weight size but only activates a subset of parameters per token, which can affect effective memory usage during inference.

MiMo-V2.5

VRAM Requirements

Compatible GPUs

MiMo-V2.5 FAQ

VRAM requirements for MiMo-V2.5 at different quantization levels
Quantization	VRAM Required
FP16	705.6 GB
Q8	352.8 GB
Q6	264.6 GB
Q4	176.4 GB