# tinfield-1-nvfp4-probe by Natapong Nitarach (Schwyter) Source: https://savrn.com/models/tinfield-1-nvfp4-probe Markdown alternate of the page above; the site index is https://savrn.com/llms.txt --- ## Runs On What it takes to serve tinfield-1-nvfp4-probe (117B 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 | 234.0 GB | 280.8 GB | 1x [MI355X](https://savrn.com/ai-index/pricing/gpus/mi355x) (288 GB) Vultr | $2.59 | [2x MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) $3.70 · [2x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $4.00 | | 8-bit | 117.0 GB | 140.4 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 · [1x MI355X](https://savrn.com/ai-index/pricing/gpus/mi355x) $2.59 | | 4-bit | 58.5 GB | 70.2 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 13, 2026. [tinfield-1-nvfp4-probe on every accelerator the SAVRN Index prices, at every precision](https://savrn.com/models/tinfield-1-nvfp4-probe/gpus) ## Model Card Public research conversion of badtheorylabs/Tinfield-1, derived from Qwen/Qwen3.8-Flash-Next. The source checkpoint was pinned to f7c714a97b71342db9e2fc04009d2704c656c80c. This Hugging Face repository contains the complete original export, including all 10 safetensors shards: - PLE embeddings, vision, MTP and other non-expert weights: retained in BF16. - The original configuration, tokenizer, processors and tensor index accompany the shards. The ple-int8-e579 Kaggle variant is a separate quantized PLE derivative. This Hugging Face checkpoint contains BF16 PLE. Load all shards and the configuration from this repository together. Conversion used NVIDIA Model-Optimizer recipe commit… Excerpt from the card by Natapong Nitarach (Schwyter), licensed other. ## Configuration Architecture Qwen4ExpForConditionalGeneration Context length (tokens) 262,144 Layers 48 Hidden size 2,560 Attention heads 24 Key/value heads 2 Head dimension 256 Vocabulary size 248,320 Experts 512 Experts active per token 10 Model type qwen4_exp Quantization modelopt ## Identity and Version Repository natnitaract/tinfield-1-nvfp4-probe Publisher Natapong Nitarach (Schwyter) Task Image and text to text Modality Image and text Library transformers Parameters 117B parameters Languages Not stated by the source Revision a0cf4f6fc6c6aec3344b2ee9a4ba9b2b67b5f9c4 First published 2026-09-24 Last updated 2026-10-05 ## Files and Weights 27 files, 186.5 GB in total. The weights are 10 files totalling 186.4 GB in safetensors. Weights10 files · 186.4 GB Configuration9 files · 32.8 MB Tokenizer4 files · 22.9 MB Documentation2 files · 6.1 KB Other1 file · 9.0 KB Repository1 file · 1.7 KB Every file | File | Type | Size | SHA-256 | | --- | --- | --- | --- | | model-00001-of-00010.safetensors | Weights | 102.4 GB | 60379ddffb28 | | model-00002-of-00010.safetensors | Weights | 10.0 GB | 8e9ad3f93d1d | | model-00003-of-00010.safetensors | Weights | 10.0 GB | f7caa06ab8af | | model-00004-of-00010.safetensors | Weights | 10.0 GB | 36c1c23148ca | | model-00005-of-00010.safetensors | Weights | 10.0 GB | 513611d4afbb | | model-00006-of-00010.safetensors | Weights | 10.0 GB | f011efc6ded7 | | model-00007-of-00010.safetensors | Weights | 10.0 GB | 1577cdd1191f | | model-00008-of-00010.safetensors | Weights | 10.0 GB | add1b2f21ccb | | model-00009-of-00010.safetensors | Weights | 8.8 GB | 9af40ba6b28a | | model-00010-of-00010.safetensors | Weights | 5.2 GB | 85bef3a7232a | | config.json | Configuration | 21.7 KB | — | | evaluation-summary.json | Configuration | 975 B | — | | generation_config.json | Configuration | 202 B | — | | hf_quant_config.json | Configuration | 18.9 KB | — | | model.safetensors.index.json | Configuration | 32.8 MB | ef8b6953d915 | | preprocessor_config.json | Configuration | 390 B | — | | processor_config.json | Configuration | 1.2 KB | — | | release-manifest.json | Configuration | 5.8 KB | — | | video_preprocessor_config.json | Configuration | 385 B | — | | LICENSE | Documentation | 3.2 KB | — | | README.md | Documentation | 2.8 KB | — | | chat_template.jinja | Other | 9.0 KB | — | | .gitattributes | Repository | 1.7 KB | — | | merges.txt | Tokenizer | 3.4 MB | — | | tokenizer.json | Tokenizer | 12.8 MB | 0997f410c57a | | tokenizer_config.json | Tokenizer | 17.9 KB | — | | vocab.json | Tokenizer | 6.7 MB | — | ## License and Download License other Access Open weights, no gate Download size 186.4 GB [Download from Natapong Nitarach (Schwyter)](https://huggingface.co/natnitaract/tinfield-1-nvfp4-probe) Released by Natapong Nitarach (Schwyter) through its official repository on Hugging Face. ## Built From - Derived from badtheorylabs/Tinfield-1 - Quantized from badtheorylabs/Tinfield-1 ## Memory Requirements | Precision | Weights in memory | | --- | --- | | As published | 186.4 GB | | 16-bit | 234.0 GB | | 8-bit | 117.0 GB | | 4-bit | 58.5 GB | Weights only, from the published parameter count; the key-value cache and runtime add to this. ## Questions About tinfield-1-nvfp4-probe ### How much GPU memory does tinfield-1-nvfp4-probe need? About 280.8 GB at 16-bit and 70.2 GB at 4-bit: the weights (117B parameters) plus a working margin. A long context needs more. ### What is the cheapest GPU to run tinfield-1-nvfp4-probe on? At 16-bit, 1x MI355X from $2.59 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists. ### What license is tinfield-1-nvfp4-probe released under? other, as its publisher declares it. Read the license text before commercial use. ### What is tinfield-1-nvfp4-probe's context length? 262,144 tokens, from the maximum position embeddings in its published configuration. ## Similar Models Model · Image and text to text ### [GLM5.3-Flash-E192-DGX-Spark](https://savrn.com/models/glm5-3-flash-e192-dgx-spark) [AutoTrust AI Lab](https://savrn.com/model-publishers/autotrust) autotrust/GLM5.3-Flash-E192-DGX-Spark is a compact build of zai-org/GLM-5.3-Flash for desktop Blackwell systems like NVIDIA DGX Spark. It's an unofficial derivative. It keeps 192 of the 288 routed experts in each layer by Neural Architecture Search (NAS), uses NVFP4 for the experts and still activates 18 B parameters per token. The weights take 124 GiB, the smallest in the family: two DGX Sparks connected by ConnectX-7 get about 62 GiB of weights per node and the most KV room of any build, and a single 180 GB Blackwell GPU (B200/GB200) runs it with 32 GiB of KV cache. The KV cache uses FP8. The repo includes a ready-to-use NVFP4 MTP draft in mtp-nvfp4/ (6.95 GB) for speculative decoding. On… Open weights mit 110.7B parameters 1,048,576 tokens vllm [View model](https://savrn.com/models/glm5-3-flash-e192-dgx-spark) Model · Image and text to text ### [Qwen3.5-122B-A10B-FP8](https://savrn.com/models/qwen3-5-122b-a10b-fp8) [Qwen](https://savrn.com/model-publishers/qwen) Over recent months, we have intensified our focus on developing foundation models that deliver exceptional utility and performance. Qwen3.5 represents a significant leap forward, integrating breakthroughs in multimodal learning, architectural efficiency, reinforcement learning scale, and global accessibility to empower developers and enterprises with unprecedented capability and efficiency. For more details, please refer to our blog post Qwen3.5. WMT24++: a harder subset of WMT24 after difficulty labeling and rebalancing; we report the averaged scores on 55 languages using XCOMET-XXL. Empty cells (--) indicate scores not yet available or not applicable. Empty cells (--) indicate scores not… Open weights apache-2.0 125.1B parameters 262,144 tokens transformers [View model](https://savrn.com/models/qwen3-5-122b-a10b-fp8) Model · Image and text to text ### [GLM5.3-Flash-E224-DGX-Spark](https://savrn.com/models/glm5-3-flash-e224-dgx-spark) [AutoTrust AI Lab](https://savrn.com/model-publishers/autotrust) autotrust/GLM5.3-Flash-E224-DGX-Spark is a compact build of zai-org/GLM-5.3-Flash for desktop Blackwell systems like NVIDIA DGX Spark. It's an unofficial derivative. It keeps 224 of the 288 routed experts in each layer by Neural Architecture Search (NAS), uses NVFP4 for the experts and still activates 18 B parameters per token. The weights take 141 GiB, which is small enough for two DGX Sparks connected by ConnectX-7 (256 GB of unified memory in total) with room left for long-context KV cache. A single 180 GB Blackwell GPU (B200/GB200) can also run it. The original GLM-5.3-Flash MTP layer ships unmodified in mtp/ (BF16) as an optional speculative-decoding draft. It gives about 1.85×… Open weights mit 127.6B parameters 1,048,576 tokens vllm [View model](https://savrn.com/models/glm5-3-flash-e224-dgx-spark) Model · Image and text to text ### [DeepSeek-V4.1-Flash-oQ2.5e-mtp](https://savrn.com/models/deepseek-v4-1-flash-oq2-5e-mtp) [Jun Kim](https://savrn.com/model-publishers/jundot) This model was quantized using oQ (oMLX v0.7.1) mixed-precision quantization with Improved oQe Quantization (jundot/omlx#4385). A mixed-precision MLX checkpoint of deepseek-ai/DeepSeek-V4.1-Flash, a multimodal mixture-of-experts model with 552B backbone parameters, a Causal Encoder-Decoder layout, Compressed Sparse Attention 2 (CSA2), Engram conditional memory and DSpark speculative decoding. It includes the language model, the Engram tables, the vision encoder, the multimodal aligner and the DSpark MTP weights. The effective average is 3.104 bits/weight excluding the Engram tables, or 3.592 bits/weight for the complete checkpoint. oQ2.5e allocates bits by measured layer sensitivity, then… Open weights mit 102.4B parameters 1,048,576 tokens mlx [View model](https://savrn.com/models/deepseek-v4-1-flash-oq2-5e-mtp) Model · Image and text to text ### [GLM5.3-Flash-E256-DGX-Spark](https://savrn.com/models/glm5-3-flash-e256-dgx-spark) [AutoTrust AI Lab](https://savrn.com/model-publishers/autotrust) autotrust/GLM5.3-Flash-E256-DGX-Spark is a compact build of zai-org/GLM-5.3-Flash for Blackwell systems, from DGX Spark clusters to a single B200. It's an unofficial derivative. It keeps 256 of the 288 routed experts in each layer by Neural Architecture Search (NAS), uses NVFP4 for the experts and still activates 18 B parameters per token. It's the largest architecture in the family, with the best Chinese, code and vision scores. The KV cache uses FP8. The repo includes a ready-to-use NVFP4 MTP draft in mtp-nvfp4/ (6.95 GB) for speculative decoding. On a single B200 it gives 1.84× single-stream decode speed (1.94× with 3 draft tokens), and output quality is unchanged. The model keeps the… Open weights mit 144.5B parameters 1,048,576 tokens vllm [View model](https://savrn.com/models/glm5-3-flash-e256-dgx-spark) Model · Image and text to text ### [Qwen3.8-Flash-Next-RAZOR-65B-A6B-NVFP4](https://savrn.com/models/qwen3-8-flash-next-razor-65b-a6b-nvfp4) [Cleyesode](https://savrn.com/model-publishers/cleyesode) Excellent 50% prune by Nickyang utilizing new RAZOR methodology. cleyesode further reduced memory requirements via nvfp4 quantization for use on Blackwell sm120 gpus. Quantization via NVIDIA/Model-Optimizer custom qwen4exp workflow: Please see post-quantization verificationreport for details. PLE isolated, full model repack for reduced memory pressure. Calibrated on curated samples from Nickyang/RazorCal: This model demonstrates very impressive retention of pre-prune / pre-quantize abilities - benchmarks in-progress. MTP is present but untested, updates will be posted as they become available. vLLM compatible via the latest branch: Example launch on 4 5060Ti 16GB GPUs (tensor parallel 4… Open weights other 88.1B parameters 262,144 tokens transformers [View model](https://savrn.com/models/qwen3-8-flash-next-razor-65b-a6b-nvfp4) ## Natapong Nitarach (Schwyter) [All models and datasets](https://savrn.com/model-publishers/natnitaract) ## Versions - [a0cf4f6fc6c6](https://savrn.com/models/tinfield-1-nvfp4-probe/versions/a0cf4f6fc6c6) · current 2026-10-05 ## Explore More - [All image and text to text models](https://savrn.com/models/tasks/image-and-text-to-text) - [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 2026-10-05. - [Hugging Face record](https://huggingface.co/natnitaract/tinfield-1-nvfp4-probe) - [How the hub is built](https://savrn.com/model-hub/methodology)