What are the hardware requirements for Inkling?

Inkling requires approximately 789.2 GB of VRAM. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Inkling and how large is it?

Inkling is developed by thinkingmachines, a model with 952.4b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does Inkling perform on standard benchmarks?

Inkling has the following published scores: LMSYS ELO 1441. Compare against reference models on this page or on the LLM Explorer leaderboards.

Inkling by thinkingmachines — VRAM 789.2GB

Name: Inkling
Author: thinkingmachines

Inkling is an open-source language model by thinkingmachines. Features: 952.4b LLM, VRAM: 789.2GB, License: apache-2.0, LLM Explorer Score: 0.62, ELO: 1441.

Audio-text-to-text Conversational Endpoints compatible Eval-results Image-text-to-text Inkling mm model Moe Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/thinkingmachines/Inkling

Inkling Benchmarks

LMSys ELO 1441 vs 1272 (so35) 13.3%

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

What is the LLM Explorer Rank (Score)

Inkling Parameters and Internals

LLM Name	Inkling
Repository 🤗	https://huggingface.co/thinkingmachines/Inkling
Model Size	952.4b
Required VRAM	789.2 GB
Updated	2026-08-08
Maintainer	thinkingmachines
Model Type	inkling_mm_model
Model Files	20.0 GB: 1-of-108 10.6 GB: 2-of-108 19.8 GB: 3-of-108 19.6 GB: 4-of-108 19.4 GB: 5-of-108 19.6 GB: 6-of-108 19.5 GB: 7-of-108 19.3 GB: 8-of-108 9.8 GB: 9-of-108 19.6 GB: 10-of-108 20.0 GB: 11-of-108 19.3 GB: 12-of-108 9.7 GB: 13-of-108 19.3 GB: 14-of-108 19.4 GB: 15-of-108 19.7 GB: 16-of-108 9.8 GB: 17-of-108 19.7 GB: 18-of-108 9.7 GB: 19-of-108 19.6 GB: 20-of-108 19.9 GB: 21-of-108 10.2 GB: 22-of-108 19.8 GB: 23-of-108 19.5 GB: 24-of-108 19.7 GB: 25-of-108 20.0 GB: 26-of-108 19.5 GB: 27-of-108 19.8 GB: 28-of-108 19.6 GB: 29-of-108 19.3 GB: 30-of-108 10.0 GB: 31-of-108 19.9 GB: 32-of-108 19.5 GB: 33-of-108 19.5 GB: 34-of-108 19.3 GB: 35-of-108 19.9 GB: 36-of-108 19.9 GB: 37-of-108 19.4 GB: 38-of-108 19.4 GB: 39-of-108 19.7 GB: 40-of-108 19.5 GB: 41-of-108 19.9 GB: 42-of-108 19.6 GB: 43-of-108 13.0 GB: 44-of-108
Model Architecture	InklingForConditionalGeneration
License	apache-2.0

Rank the Inkling capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Quantized Models of the Inkling

Model	Likes	Downloads	VRAM
Inkling Mlx 2bit	6	3045	225 GB
Inkling Mlx 4bit	4	551	391 GB

Best Alternatives to Inkling

Best Alternatives	Context / RAM	Downloads	Likes
Vedika	0K / 789.2 GB	12	1