What are the hardware requirements for Hy3 Preview?

Hy3 Preview requires approximately 221.4 GB of VRAM and supports a context window of 256K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Hy3 Preview and how large is it?

Hy3 Preview is developed by tencent, a model with 298.8b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does Hy3 Preview perform on standard benchmarks?

Hy3 Preview has the following published scores: LMSYS ELO 1412. Compare against reference models on this page or on the LLM Explorer leaderboards.

Hy3 Preview by tencent — VRAM 221.4GB, 256K context

Name: Hy3 Preview
Author: tencent

Hy3 Preview is an open-source language model by tencent. Features: 298.8b LLM, VRAM: 221.4GB, Context: 256K, License: other, LLM Explorer Score: 0.5, ELO: 1412.

Conversational Endpoints compatible Eval-results Hy v3 Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/tencent/Hy3-preview

Hy3 Preview Benchmarks

LMSys ELO 1412 vs 1272 (so35) 11%

^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)

Hy3 Preview Parameters and Internals

LLM Name	Hy3 Preview
Repository 🤗	https://huggingface.co/tencent/Hy3-preview
Model Size	298.8b
Required VRAM	221.4 GB
Updated	2026-07-19
Maintainer	tencent
Model Type	hy_v3
Model Files	5.4 GB: 1-of-112 5.4 GB: 2-of-112 5.4 GB: 3-of-112 5.4 GB: 4-of-112 5.4 GB: 5-of-112 5.4 GB: 6-of-112 5.4 GB: 7-of-112 5.4 GB: 8-of-112 5.4 GB: 9-of-112 5.4 GB: 10-of-112 5.4 GB: 11-of-112 5.4 GB: 12-of-112 5.4 GB: 13-of-112 5.4 GB: 14-of-112 5.4 GB: 15-of-112 5.4 GB: 16-of-112 5.4 GB: 17-of-112 5.4 GB: 18-of-112 5.4 GB: 19-of-112 5.4 GB: 20-of-112 5.4 GB: 21-of-112 5.4 GB: 22-of-112 5.4 GB: 23-of-112 5.4 GB: 24-of-112 5.4 GB: 25-of-112 5.4 GB: 26-of-112 5.4 GB: 27-of-112 5.4 GB: 28-of-112 5.4 GB: 29-of-112 5.4 GB: 30-of-112 5.4 GB: 31-of-112 5.4 GB: 32-of-112 5.4 GB: 33-of-112 5.4 GB: 34-of-112 5.4 GB: 35-of-112 5.4 GB: 36-of-112 5.4 GB: 37-of-112 5.4 GB: 38-of-112 5.4 GB: 39-of-112 5.4 GB: 40-of-112 5.4 GB: 41-of-112
Model Architecture	HYV3ForCausalLM
License	other
Context Length	262144
Model Max Length	262144
Transformers Version	5.6.0
Vocabulary Size	120832

Rank the Hy3 Preview 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 Hy3 Preview

Model	Likes	Downloads	VRAM
Hy3 Preview 4bit	3	1563	166 GB
Hy3 Preview 8bit	1	141	235 GB
Hy3 Preview 6bit	1	93	227 GB

Best Alternatives to Hy3 Preview

Best Alternatives	Context / RAM	Downloads	Likes
Hy3	256K / 258.6 GB	18600	873
Hy3 FP8	256K / 129.6 GB	11017	54
Hy3 Preview Base	256K / 265.5 GB	618	24

Note: green Score (e.g. "73.2") means that the model is better than tencent/Hy3-preview.