Can you run Whisper large-v3 locally?

Whisper family · November 2023 · 1.54B parameters · Hugging Face model card

How much VRAM does Whisper large-v3 need at each quantization?

Whisper large-v3 needs 2.0 GB of VRAM at Q4_K_M. The table below lists weights-only size and total VRAM including overhead for each common quantization level including runtime overhead.

Encoder-decoder transformer — audio has no growing KV cache (30-second windows). Table adds a 1 GB runtime overhead (Mel spectrogram, beam search buffers).

Method: weights = parameters × bits-per-weight (Q4_K_M ≈ 4.8, Q5_K_M ≈ 5.7, Q6_K ≈ 6.6, Q8_0 ≈ 8.5, FP16 = 16), plus 10% loading overhead, plus KV cache from the verified architecture config (not applicable to this architecture). Source: Params 1.54B, d_model 1280, 32 encoder + 32 decoder layers, 128 Mel bins — HF config.json + model card (openai/whisper-large-v3).

Which GPUs can run Whisper large-v3 locally?

At Q4_K_M with a 4k context, Whisper large-v3 needs 2.0 GB of VRAM. The lists below are computed live from our GPU database and grouped by how much headroom the card has. Disclosure: CompareAIHardware.com participates in the Amazon Associates program and earns from qualifying purchases through links on this page. Affiliate relationships do not influence our recommendations.

Runs Whisper large-v3 comfortably (25%+ VRAM headroom)

Can a Mac run Whisper large-v3?

Yes — these Apple Silicon machines fit Whisper large-v3 at Q4_K_M, since macOS lets the GPU use about 75% of unified memory. Generation speed is bound by memory bandwidth, so the GB/s column matters as much as capacity.

Can Whisper large-v3 run on mini PCs, Jetson, or NPU devices?

These edge and NPU devices from our database have enough memory for Whisper large-v3 at Q4_K_M. Their memory bandwidth is far below discrete GPUs, so expect a fraction of desktop generation speed.

Which pre-built systems can run Whisper large-v3?

These mini PCs and workstations from our database fit Whisper large-v3 at Q4_K_M. Usable-memory figures are conservative: Windows shares about half of system RAM with the GPU by default (Linux can expose more), macOS lets Apple Silicon GPUs use about 75% of unified memory, and Linux unified-memory systems such as GB10 expose roughly 90%. Generation speed is bound by memory bandwidth, so compare the GB/s column before buying.

Frequently asked questions

Related guides

• Best GPU for Whisper
• Best GPU Under $500
• VRAM Calculator