Qwen3-4B-Instruct-2507

Alibaba · 4.0B · ctx 256K · Apache 2.0 · HuggingFace

Qwen3-4B-Instruct-2507 (Alibaba) is a 4.0B parameters open-weight model tracked on AI Hashrate. At 8K context we estimate roughly 4.0 GB VRAM for Q4 (weights 2.2 GB + KV 0.8 GB + 1 GB overhead) and 9.8 GB for FP16. About 50 GPUs/accelerators in our catalog fully fit this model at Q4 under that context assumption. Fastest Q4 configs: B200 192GB SXM ≈ 1272.7 tok/s (estimated); MI300X 192GB ≈ 843.2 tok/s (estimated); H200 141GB SXM5 ≈ 763.6 tok/s (estimated); H100 80GB SXM5 ≈ 533.0 tok/s (estimated); Gaudi 3 128GB ≈ 521.2 tok/s (estimated). Use the table below for VRAM fit and tok/s per dollar (list MSRP). Estimates follow memory-bandwidth math; measured rows override when present. See methodology for details. Methodology.

100 hardware configs — sorted by tok/s. Default fit context 8K.