Qwen3.6 NVFP4 Quantization Speeds Up
danielhanchen · reddit · 2026-07-10
Unsloth released an NVFP4 quantization scheme for Qwen3.6, claiming that without losing precision, it boosted inference speeds for the 27B model to 2.5x that of NVIDIA NVFP4, and the 35B-A3B to 1.56x–1.79x.
The post also provides several benchmark comparisons, including MMLU-Pro, GPQA, AIME 2025, and results against BF16 / FP8 / NVIDIA NVFP4. The author mentioned adding FP8 KV Cache calibration, which automatically doubles the context length to 2x. The text distinguishes between two versions of the 35B model: one optimized for speed (NVFP4-Fast), and another making a slight compromise between speed and precision.
The relevant quantized models are available on Hugging Face, with a more comprehensive analysis and benchmarks in their blog post.
Related event: Unsloth Brings Faster NVFP4 Quantization to Qwen3.6(3 posts)→
More from Infra
- Anthropic S-1 reveals 48 cents of compute cost per dollar of Q2 revenue — nmasc_ · 2026-10-26
- Persistent AI joins UR network to route private AI inference with encrypted, IP-hidden traffic — 0xSammy · 2026-10-26
- Anthropic's disclosed compute commitments: $114.5B Google, $161.2B Broadcom top the list — BenBajarin · 2026-10-26
- SpaceX strikes $40B deal with Apollo Global to lease AI chips and data center hardware — Polymarket · 2026-10-26
- SpaceX AI revenue up 830%, Claude designs 318K antivenom proteins in 64 hours — PeterDiamandis · 2026-10-26
- Meta plans up to $95 billion stock sale to fund AI buildout — The Decoder · 2026-10-26