# Qwen3-32B-NVFP4-W4A4 by Cezar: Open-Weight Model Source: https://savrn.com/models/qwen3-32b-nvfp4-w4a4 Markdown alternate of the page above; the site index is https://savrn.com/llms.txt --- ## Runs On What it takes to serve Qwen3-32B-NVFP4-W4A4 (19.1B parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them. | Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits | | --- | --- | --- | --- | --- | --- | | 16-bit | 38.2 GB | 45.9 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 8-bit | 19.1 GB | 22.9 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 4-bit | 9.6 GB | 11.5 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the [SAVRN Index](https://savrn.com/ai-index/pricing/gpus), read Oct 9, 2026. [Qwen3-32B-NVFP4-W4A4 on every accelerator the SAVRN Index prices, at every precision](https://savrn.com/models/qwen3-32b-nvfp4-w4a4/gpus) ## Model Card By Cezar, published under apache-2.0, revision a274b18165ad. Benchmark code, raw data and results: [github.com/cezarc1/decodebench](https://github.com/cezarc1/decodebench) (see [RESULTS.md](https://github.com/cezarc1/decodebench/blob/main/RESULTS.md)). NVFP4 W4A4 quantization of [Qwen/Qwen3-32B](https://savrn.com/models/qwen3-32b) (revision 9216db5781bf21249d130ec9da846c4624c16137, BF16). It was made for a controlled NVFP4-vs-MXFP4 decode benchmark on NVIDIA B200 (fp4bench), not as a general-purpose release: the NVFP4 and MXFP4 checkpoints share the model, the tool, the recipe, the calibration data and the layer coverage, and differ only in the format. | Weights | NVFP4: FP4 E2M1 values in 16-element blocks, E4M3 block scales plus one FP32 per-tensor global scale, 4.5 bits per weight | | --- | --- | | Activations | 4-bit, quantized per 16-element block at runtime, with a static per-tensor activation scale (input_global_scale) calibrated on the data below | | Coverage | all Linear layers except lm_head (448 modules); embeddings, norms and lm_head stay BF16 | | KV cache | BF16 at serve time (--kv-cache-dtype bfloat16); the checkpoint has no kv_cache_scheme | ### How it was made [Read the full model card (341 words)](https://savrn.com/models/qwen3-32b-nvfp4-w4a4/card) ## Configuration Architecture Qwen3ForCausalLM Context length (tokens) 40,960 Layers 64 Hidden size 5,120 Feed-forward size 25,600 Attention heads 64 Key/value heads 8 Head dimension 128 Vocabulary size 151,936 RoPE base 1,000,000 Stored precision bfloat16 Model type qwen3 Quantization compressed-tensors ## Identity and Version Repository ggamecrazy/Qwen3-32B-NVFP4-W4A4 Publisher Cezar Task Text generation Modality Text Library Not stated by the source Parameters 19.1B parameters Languages Not stated by the source Revision a274b18165ad44a9627ff8aee7626e5fb47ee2f6 First published 2026-10-05 Last updated 2026-10-09 ## Files and Weights 14 files, 20.7 GB in total. The weights are 2 files totalling 20.7 GB in safetensors. Weights2 files · 20.7 GB Configuration5 files · 205.7 KB Tokenizer4 files · 15.9 MB Documentation2 files · 14.4 KB Repository1 file · 1.6 KB Every file | File | Type | Size | SHA-256 | | --- | --- | --- | --- | | model-00001-of-00002.safetensors | Weights | 19.1 GB | ec718c8798ae | | model-00002-of-00002.safetensors | Weights | 1.6 GB | fbd6a7104a88 | | config.json | Configuration | 2.1 KB | — | | fp4bench_quant.json | Configuration | 18.5 KB | — | | generation_config.json | Configuration | 239 B | — | | model.safetensors.index.json | Configuration | 184.6 KB | — | | recipe.yaml | Configuration | 208 B | — | | LICENSE | Documentation | 11.3 KB | — | | README.md | Documentation | 3.1 KB | — | | .gitattributes | Repository | 1.6 KB | — | | merges.txt | Tokenizer | 1.7 MB | — | | tokenizer.json | Tokenizer | 11.4 MB | aeb13307a71a | | tokenizer_config.json | Tokenizer | 9.7 KB | — | | vocab.json | Tokenizer | 2.8 MB | — | ## License and Download License apache-2.0 Access Open weights, no gate Download size 20.7 GB [Download from Cezar](https://huggingface.co/ggamecrazy/Qwen3-32B-NVFP4-W4A4) Released by Cezar through its official repository on Hugging Face. [Read the license](https://www.apache.org/licenses/LICENSE-2.0). ## Built From - Derived from [Qwen/Qwen3-32B](https://savrn.com/models/qwen3-32b) - Quantized from [Qwen/Qwen3-32B](https://savrn.com/models/qwen3-32b) ## Memory Requirements | Precision | Weights in memory | | --- | --- | | As published | 20.7 GB | | 16-bit | 38.2 GB | | 8-bit | 19.1 GB | | 4-bit | 9.6 GB | Weights only, from the published parameter count; the key-value cache and runtime add to this. ## Questions About Qwen3-32B-NVFP4-W4A4 ### How much GPU memory does Qwen3-32B-NVFP4-W4A4 need? About 45.9 GB at 16-bit and 11.5 GB at 4-bit: the weights (19.1B parameters) plus a working margin. A long context needs more. ### What is the cheapest GPU to run Qwen3-32B-NVFP4-W4A4 on? At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists. ### Can I use Qwen3-32B-NVFP4-W4A4 commercially? Yes. Qwen3-32B-NVFP4-W4A4 is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors. ### What is Qwen3-32B-NVFP4-W4A4's context length? 40,960 tokens, from the maximum position embeddings in its published configuration. ## Similar Models Model · Text generation ### [K2-Horizon-MoVA-36B-A4B-NSLM-p4mx-q4v](https://savrn.com/models/k2-horizon-mova-36b-a4b-nslm-p4mx-q4v) [Matthew P Barnson](https://savrn.com/model-publishers/txgsync) An experimental NanoSeedLM (SeedLM) compression of IFM/K2-Horizon-MoVA-36B-A4B for Apple Silicon Macs with 32 GB of unified memory. SeedLM keeps each block of 8 weights as a 16-bit seed of a linear feedback shift register (LFSR), 4 coefficients and an exponent. The GPU makes the weights again from the seed at run time. The paper used an FPGA for this. This model uses Metal kernels on the Mac GPU. This is a research proof of concept. It is not a replacement for a calibrated quantization. - Six safetensors shards (22.76 GB in total, at most 4.5 GB each) and model.safetensors.index.json. The other files are config.json, the tokenizer files of the base model and nanoseedlmk2.py, the MLX loader.… Open weights apache-2.0 19.2B parameters 524,288 tokens mlx [View model](https://savrn.com/models/k2-horizon-mova-36b-a4b-nslm-p4mx-q4v) Model · Text generation ### [Ornith-1.5-35B-A3B-NVFP4](https://savrn.com/models/ornith-1-5-35b-a3b-nvfp4) [Ornith](https://savrn.com/model-publishers/ornith-ai) Chirp Chirp! We are introducing Ornith-1.5, a major step toward building foundation models through end-to-end self-improvement. Ornith-1.5 extends Ornith-1.0 (which was developed on top of Qwen3.5 and Gemma4 with additional continued pretraining, mid-training, and post-training) by expanding the self-improvement loop from scaffold and rollout optimization to jointly optimizing task generation, scaffold construction, and solution rollouts. Rather than relying on a fixed set of human-curated tasks and manually designed harnesses, Ornith-1.5 continuously generates new training tasks, discovers effective strategies for solving them, and improves the policy through reinforcement learning. For… Open weights mit 19.5B parameters 262,144 tokens transformers [View model](https://savrn.com/models/ornith-1-5-35b-a3b-nvfp4) Model · Text generation ### [Qwen3.6-35B-A3B-NVFP4](https://savrn.com/models/qwen3-6-35b-a3b-nvfp4) [NVIDIA](https://savrn.com/model-publishers/nvidia) The NVIDIA Qwen3.6-35B-A3B-NVFP4 model is the quantized version of Alibaba's Qwen3.6-35B-A3B model, which is an auto-regressive language model that uses an optimized transformer architecture. For more information, please check here. The NVIDIA Qwen3.6-35B-A3B-NVFP4 model is quantized with Model Optimizer. This model is ready for commercial/non-commercial use. This model is not owned or developed by NVIDIA. This model has been developed and built to a third-party’s requirements for this application and use case; see link to Non-NVIDIA (Qwen3.6-35B-A3B) Model Card from Alibaba. Global Developers looking to take off-the-shelf, pre-quantized models for deployment in AI Agent systems, chatbots… Open weights apache-2.0 18.7B parameters 262,144 tokens Model Optimizer [View model](https://savrn.com/models/qwen3-6-35b-a3b-nvfp4) Model · Text generation ### [veriloop-coder-e2-nvfp4](https://savrn.com/models/veriloop-coder-e2-nvfp4) [Rodrigo Ramos Da Silveira](https://savrn.com/model-publishers/rodrigoramosrs) Apache-2.0 This repository contains an NVFP4 quantization of VeriLoop E2, an open 27B post-trained model built on Qwen3.8-27B for code, mathematics, and physics. Its core reasoning discipline is VeriLoop-Governed Recurrence (VGR): candidate states are recursively proposed, externally checked, and retained only when the protected evidence state improves without regression. Quantized by Rodrigo Ramos. Produced with NVIDIA Model Optimizer using the canonical NVFP4W4A4WEIGHTLOCALHESSIANCFG recipe: static per-block 4-bit weights (group size 16) + dynamic 4-bit activations, FP8 attention, local-Hessian calibration with MSE and fp8 scale sweep. The linearattn (Gated Delta Net) blocks and the… Open weights apache-2.0 18.3B parameters 262,144 tokens transformers [View model](https://savrn.com/models/veriloop-coder-e2-nvfp4) Model · Text generation ### [NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4](https://savrn.com/models/nvidia-nemotron-3-nano-30b-a3b-nvfp4) [NVIDIA](https://savrn.com/model-publishers/nvidia) September 2025 \- December 2025 The post-training data has a cutoff date of November 28, 2025\. The pre-training data has a cutoff date of June 25, 2025\. Nemotron-Nano-3-30B-A3B-NVFP4 is a quantized version of Nemotron-Nano-3-30B-A3B and is a large language model (LLM) trained from scratch by NVIDIA, and designed as a unified model for both reasoning and non-reasoning tasks. It responds to user queries and tasks by first generating a reasoning trace and then concluding with a final response. The model's reasoning capabilities can be configured through a flag in the chat template. If the user prefers the model to provide its final answer without intermediate reasoning traces, it can be… Open weights other 18.2B parameters 262,144 tokens transformers [View model](https://savrn.com/models/nvidia-nemotron-3-nano-30b-a3b-nvfp4) Model · Text generation ### [NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4](https://savrn.com/models/nvidia-nemotron-3-5-lightning-30b-a3b-nvfp4) [NVIDIA](https://savrn.com/model-publishers/nvidia) The pre-training data has a cutoff date of September 2025. The post-training data has a cutoff date of May 2026. NVIDIA Nemotron™ is a family of open models with open weights, training data, and recipes, delivering leading efficiency and accuracy for building specialized AI agents. NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4 is a large language model (LLM) trained by NVIDIA. The model employs a hybrid Mixture-of-Experts architecture, utilizing interleaved Mamba-2 and MoE layers, along with select Attention layers. The Lightning 3.5 model is released alongside a number of speculative decoding methods for faster text generation. The model has 3B active parameters and 30B parameters in total.… Open weights other 17.8B parameters 1,048,576 tokens transformers [View model](https://savrn.com/models/nvidia-nemotron-3-5-lightning-30b-a3b-nvfp4) ## Cezar [All models and datasets](https://savrn.com/model-publishers/ggamecrazy) ## Versions - [a274b18165ad](https://savrn.com/models/qwen3-32b-nvfp4-w4a4/versions/a274b18165ad) · current 2026-10-09 ## Explore More - [All text generation models](https://savrn.com/models/tasks/text-generation) - [All models under apache-2.0](https://savrn.com/models/licenses/apache-2-0) - [Model comparisons](https://savrn.com/models/comparisons) - [The model directory](https://savrn.com/models) - [Open model prices by host](https://savrn.com/ai-index/pricing/open-models) ## Source - Repository metadata, read 2026-10-09. - [Hugging Face record](https://huggingface.co/ggamecrazy/Qwen3-32B-NVFP4-W4A4) - [How the hub is built](https://savrn.com/model-hub/methodology)