Phi-3 mini VRAM and GPU Requirements

Phi-3 mini needs about 3.75 GB at Q4_K_M with 4,096 tokens.

Which GPUs fit

VRAM by quantization

Totals at 4,096 tokens. Q4_K_M is the usual balance of size and quality. FP16 weights are the parameter count times 2 bytes. The FP16 column in the table is the full estimate: weights, cache, and 1 GB.

Context length

Q4_K_M total as the context changes. Cache is the part that grows.

• 2,048 3.75 GB
• 4,096 3.75 GB

Phi-3 mini hardware notes

• Unified memory counts 75 percent of the listed pool as usable. Discrete cards use the full listed memory. The memory column still shows the listed figure.
• Last updated 2025-12-10.
• Weights, cache, and the 1 GB overhead are estimates. How this is calculated.
• Architecture numbers and the parameter count come from the public model record, checked 2026-10-10.

Phi-3 mini VRAM FAQ

How much VRAM does Phi-3 mini need at Q4_K_M?

3.75 GB at 4,096 tokens. Weights are 2.00 GB, the cache is 0.75 GB, and overhead is 1 GB.

Does a 24 GB GPU fit Phi-3 mini at Q4_K_M?

GeForce RTX 3090 24GB is Fits, 20.25 GB left. The configured maximum is 4,096 tokens, which is under 8,192, so the quant table uses 4,096 tokens.

What context length is configured for Phi-3 mini?

The configured context length is 4,096 tokens. The configured maximum is 4,096 tokens, which is under 8,192, so the quant table uses 4,096 tokens.

Which source supplies the parameter count for Phi-3 mini?

3.82B parameters, from the Hugging Face model record for microsoft/Phi-3-mini-4k-instruct.

Related models