LLM EXPLORER 58,608 MODELS INDEXED

Inkling NVFP4 by thinkingmachines

By thinkingmachines · 0 downloads
HF Repo ↗ Copy URL

Inkling NVFP4 is an open-source language model by thinkingmachines. Features: 552.8b LLM, VRAM: 581.4GB, License: apache-2.0, LLM Explorer Score: 0.41, ELO: 1442.

  8-bit   Audio-text-to-text   Conversational   Image-text-to-text   Inkling mm model   Moe   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/thinkingmachines/Inkling-NVFP4 

Inkling NVFP4 Benchmarks

LMSys ELO 1442 vs 1272 (so35) 13.4%
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

Inkling NVFP4 Parameters and Internals

LLM NameInkling NVFP4
Repository 🤗https://huggingface.co/thinkingmachines/Inkling-NVFP4 
Model Size552.8b
Required VRAM581.4 GB
Updated2026-07-16
Maintainerthinkingmachines
Model Typeinkling_mm_model
Model Files  19.1 GB: 1-of-33   20.0 GB: 2-of-33   19.2 GB: 3-of-33   15.7 GB: 4-of-33   17.9 GB: 5-of-33   19.9 GB: 6-of-33   19.2 GB: 7-of-33   20.0 GB: 8-of-33   18.7 GB: 9-of-33   20.0 GB: 10-of-33   19.5 GB: 11-of-33   15.6 GB: 12-of-33   18.1 GB: 13-of-33   18.0 GB: 14-of-33   19.8 GB: 15-of-33   19.6 GB: 16-of-33   18.3 GB: 17-of-33   19.5 GB: 18-of-33   15.5 GB: 19-of-33   19.3 GB: 20-of-33   16.6 GB: 21-of-33   17.8 GB: 22-of-33   17.8 GB: 23-of-33   16.0 GB: 24-of-33   19.4 GB: 25-of-33   19.1 GB: 26-of-33   18.3 GB: 27-of-33   19.6 GB: 28-of-33   19.9 GB: 29-of-33   1.2 GB: 30-of-33   19.4 GB: 31-of-33   18.6 GB: 32-of-33   4.8 GB: 33-of-33   10.5 GB
Model ArchitectureInklingForConditionalGeneration
Licenseapache-2.0
Tokenizer ClassPreTrainedTokenizerFast

Quantized Models of the Inkling NVFP4

Model
Likes
Downloads
VRAM
Inkling NVFP4 Mlx 4bit11119373 GB

Best Alternatives to Inkling NVFP4

Best Alternatives
Context / RAM
Downloads
Likes
Inkling NVFP40K / 581.4 GB28811