Hy3
tencentothertext

Hy3

Updated Jul 16, 2026
Parameters
298.8B
Context
262,144
License
other
Updated
Jul 16, 2026

Intelligence benchmarks

Artificial Analysis indexes - compared with the best open and proprietary models

Intelligence

41.2

AA Index

Coding

58.8

AA Index

Agentic

30.7

AA Index

Intelligence Index - Hy3 vs. the field

Best open-weight models (you can run locally) and leading proprietary models for context.

Claude Fable 5 (with fallback)
59.9
closed
GPT-5.6 Sol (max)
58.9
closed
Claude Opus 4.8 (max)
55.7
closed
GPT-5.6 Terra (max)
55
closed
GPT-5.5 (xhigh)
54.8
closed
GLM-5.2
51.1
open
MiniMax-M3
44.4
open
Hy3
41.2
open

Coding Index comparison

GPT-5.6 Sol (xhigh)
78.3
closed
GPT-5.6 Terra (max)
76.7
closed
Claude Fable 5 (with fallback)
76.5
closed
GPT-5.5 (xhigh)
74.9
closed
Claude Opus 4.8 (max)
74.3
closed
GLM-5.2
68.8
open
Kimi K2.6
61.8
open
Hy3
58.8
open

Agentic Index comparison

GPT-5.6 Sol (max)
54
closed
Claude Fable 5 (with fallback)
52.8
closed
GPT-5.6 Terra (max)
47.4
closed
Claude Opus 4.8 (max)
47.2
closed
Claude Sonnet 5 (max)
46.7
closed
GLM-5.2
43.1
open
DeepSeek V4 Pro
36.4
open
Hy3
30.7
open

Benchmark data from Artificial Analysis · updated 2026-07-16.

Standard benchmarks

Performance across standard evaluations

BenchmarkScore
GPQA89.7

Will it run on your hardware?

Pick your GPU memory - see which quantizations fit, and the cheapest card for the rest

Too big for 24 GB at any quant
0 of 4 quantizations fit Hy3 with real runtime overhead.
Q5_K_M
212 GB
Q8_0
320 GB
FP16
598 GB
fits tight too big

Need an exact figure for your context length? Use the VRAM calculator.

Run it locally

Copy-paste - running in under a minute

vLLMOpenAI-compatible API
vllm serve tencent/Hy3-preview

New to this? Start with Ollama · serve to many users with vLLM.

Deep dive

Notes, sources, and the full write-up

Hy3 is a 298.8B-parameter other model from Tencent. It scores 41.2 on the Artificial Analysis Intelligence Index (coding 58.8). At Q4_K_M it needs roughly 173 GB of VRAM, placing it in the 48 GB+ / multi-GPU hardware tier.

Benchmarks

Artificial Analysis Intelligence Index - Hy3 vs. leading closed models:

ModelIntelligenceCodingGPQA
Hy341.258.889.7
Claude Fable 5 (with fallback)59.976.592.6
GPT-5.6 Sol (max)58.977.494.1
Claude Opus 4.8 (max)55.774.392
GPT-5.6 Terra (max)5576.792.5
GPT-5.5 (xhigh)54.874.993.5

Source: Artificial Analysis (2026-07-16).

Popularity

Hy3 has 78,998 downloads in the last month on HuggingFace and 296 likes.

Frequently asked

Quick answers to common questions

How much VRAM does Hy3 need?

Hy3 with 298.8B parameters needs approximately 173 GB at Q4_K_M quantization. Use our VRAM calculator for an exact estimate.

Is Hy3 better than other tencent models?

Hy3 has 298.8B parameters with 262,144 context - a strong choice for general use.

What license is Hy3 under?

Hy3 is released under the other license, making it suitable for most commercial and personal projects.

What hardware runs Hy3 well?

With 298.8B parameters, Hy3 requires adequate VRAM. High-end GPUs like the RTX 4090 (24GB), RTX 5090 (32GB), or Mac Studio with unified memory are good options. Check our hardware directory for specific recommendations.

What is the best quantization for Hy3?

Q4_K_M is the recommended sweet spot - ~98% of FP16 quality at ~27% of the size. Q5_K_M (~212 GB) is an option if you have spare VRAM. Use our VRAM calculator to compare.

How long can Hy3's context window handle?

Hy3 supports a 262,144-token context window - enough for very long documents, codebases, or multi-turn conversations. Real-world usable context may vary by implementation.

What models compete with Hy3?

Hy3 competes with other 149B–448B. Browse our model directory for comparisons, benchmarks, and community reviews to find the best fit.

Nearby options

Similar models and compatible hardware by spec

Comments coming soon

Configure NEXT_PUBLIC_GISCUS_REPO_ID and NEXT_PUBLIC_GISCUS_CATEGORY_ID at giscus.app to enable.