Are the Hy3 weights open and can I run it locally?

Yes, the weights are published on Hugging Face under tencent/Hy3 and the model card covers deployment with vLLM and SGLang, as well as finetuning and quantisation. At 295B total parameters you will need substantial hardware — check the VRAM guide before committing.

What does Hy3 cost and what is the context window?

Pricing is $0.08 per million input tokens and $0.33 per million output tokens, with a 256K token context window.

How does Hy3 differ from other large MoE models?

Hy3 activates only 21B of its 295B parameters per forward pass, keeping inference costs lower than dense models of similar quality. Its distinguishing focus is production reliability: reduced hallucination rates, stable tool-calling across agent frameworks, and configurable reasoning effort rather than always-on chain-of-thought.

▲

33.9%

this week

Hy3

9181°

#llm #tencent #reasoning #coding #agents #long context

Run this model

Offers / Ads

OpenRouter Terminal / API

Pay per token

Chat with this model in the browser or call it from your code with one API key.

Tencent | 262,144 token context | $0.13/M input, $0.53/M output. Hy3 is a 295B-parameter Mixture-of-Experts model from Tencent (21B active, 192 experts with top-8 routing) built for reasoning, agentic workflows, and real-world production use. It supports a configurable reasoning effort: a direct no-think mode by default, plus low and high chain-of-thought modes for complex math, coding, and multi-step problems. With a 256K context window, Hy3 targets long-horizon tasks, including improved coreference resolution, multi-turn constraint tracking, and stable tool-calling that generalizes across…Tencent | 262,144 token context | $0.13/M input, $0.53/M output. Hy3 is a 295B-parameter Mixture-of-Experts model from Tencent (21B active, 192 experts with top-8 routing) built for reasoning, agentic workflows, and real-world production use. It supports a configurable reasoning effort: a direct no-think mode by default, plus low and high chain-of-thought modes for complex math, coding, and multi-step problems. With a 256K context window, Hy3 targets long-horizon tasks, including improved coreference resolution, multi-turn constraint tracking, and stable tool-calling that generalizes across agent scaffoldings. Tencent positions it as a reliable, cost-effective option across coding, document processing, financial analysis, game development, and frontend design, with a strong emphasis on grounded, anti-hallucination behavior that answers when grounded and flags when evidence is missing rather than fabricating.

Guides for this model

How much VRAM do I actually need for local AI?

What is a GGUF file? Quantized models explained (Q4, Q8, fp8, NF4)

Open weights vs open source vs API-only: what you are actually allowed to do

All guides

Captain's notes

Written by Captain, the house editor

Notes from 2026-10-13

Hy3 is a 295B Mixture-of-Experts model from Tencent, with 21B parameters active at inference time and a 256K context window. It ships with switchable reasoning depth — no-think by default, chain-of-thought on demand — and was built with production reliability in mind: stable tool-calling, reduced hallucination, and consistent multi-turn constraint tracking. At $0.08/M input and $0.33/M output, it targets teams who need a capable, grounded workhorse without flagship-model pricing.

Known for

Configurable reasoning depth: silent by default, full chain-of-thought when you need it
Anti-hallucination design: answers when grounded, flags gaps rather than fabricating
Stable tool-calling that generalises across agent scaffoldings like Cline and KiloCode
256K context window suited to long documents and multi-turn constraint tracking
Open weights with support for vLLM, SGLang, finetuning, and RL post-training

Watch out for

295B total parameters means serious hardware to run locally — plan for multi-GPU or quantised deployment.
Weights are open but check the licence terms before commercial or derivative use.
Tencent has not published a standalone API; availability depends on your inference provider.

Best for

Prompts to try

Multi-step agentic plan

You are building a CI/CD pipeline for a Python monorepo with three services. Each service has its own tests and Docker image. Outline the pipeline stages, the tool calls you would make at each stage, and how you would handle a failure in service B without blocking service C from deploying.

Grounded financial analysis

Here is a company's income statement and balance sheet for the last two fiscal years. Identify the three most significant changes in financial health, explain what each change likely signals, and flag any figures where the data is ambiguous or where you would need additional context before drawing a conclusion.

Hard reasoning with evidence

A train leaves city A at 08:00 travelling at 90 km/h. A second train leaves city B, 360 km away, at 09:00 travelling toward city A at 120 km/h. At what time do they meet, and how far from city A? Show each step of your reasoning before giving the answer.

Compare with

Hy4 preview
Open-weights model
1,048,576 token context · $0.83/M input, $2.50/M output
Hy3 preview
Open-weights model
262,144 token context · $0.18/M input, $0.60/M output
Hunyuan A13B Instruct
Open-weights model
131,072 token context · $0.14/M input, $0.57/M output
GPT-6 Luna
API-only model
1,050,000 token context · $0.10/M input, $0.50/M output

More text models from tencent

Questions people ask

Are the Hy3 weights open and can I run it locally?: Yes, the weights are published on Hugging Face under tencent/Hy3 and the model card covers deployment with vLLM and SGLang, as well as finetuning and quantisation. At 295B total parameters you will need substantial hardware — check the VRAM guide before committing.
What does Hy3 cost and what is the context window?: Pricing is $0.08 per million input tokens and $0.33 per million output tokens, with a 256K token context window.
How does Hy3 differ from other large MoE models?: Hy3 activates only 21B of its 295B parameters per forward pass, keeping inference costs lower than dense models of similar quality. Its distinguishing focus is production reliability: reduced hallucination rates, stable tool-calling across agent frameworks, and configurable reasoning effort rather than always-on chain-of-thought.

Captain's verdict
A serious open-weights production model — grounded, agent-ready, and priced well below the flagship tier it competes with.

Model card on Hugging Face

Captain writes these notes from the model's own description and model card. Prices and context sizes are the provider's at the time of writing and can change.

Ranking badge

Updated Oct 22, 2026

Made this model? Show this badge on your page.

Paste it into your Hugging Face card or Civitai description.

Change colors / personalize / get code

Also ranked: #13 Hottest Text / LLM model · #13 Hottest Text / LLM text model

Not this creator?

Get your own badge

Hy3 Popularity Trend

Heat score since 6 Jul 2026

Heat score

9,181°

This week

▲

33.9%

9,142° then, 9,181° now

This month

▲

315%

8,992° then, 9,181° now

Past year

n/a

Wasn't released/tracked

7,000°

Floor

The scale runs from 0° to 9,999°. Everything below 7,000° is kept for models no source reports on yet, which carry a stand-in number there so lists stay complete. 7,000° is where measured data begins: a model with real data but almost no downloads sits at the floor, and the measurements fill the degrees above it. The floor says nothing about the model; only the degrees above it do, and they are what separates one model from another.

+1,828°

OpenRouter tokens

12,363,366,372,545 tokens this model processed through OpenRouter in the last 30 days, prompts and answers together. Every 10,000,000 tokens count as one download in the download term, 1,236,336 in all; the last day's tokens join the adoption rate. Source: OpenRouter (openrouter.ai/rankings).

+353°

Adoption rate

About 50,937 downloads a day, measured between two readings of the sources, never guessed from a lifetime average. Speed weighs a quarter of the download term, so a fresh surge shows without letting a one-day spike outrank years of use.

Show detailed breakdown

Every number on this site can be checked. This model's heat score is 9,181°, and here it is built one ingredient at a time, with the running total after each: the floor, the degrees each source adds, the bonus for the kind of model and what the age of its weights takes away. The colors are the bands of the chart above. The glossary at the end defines every term.

1
Start at the floor.
Why 7,000° and not zero: the scale runs from 0° to 9,999°, and the part below 7,000° is reserved for models no source reports on yet, which carry a stand-in number there so the lists stay complete and never outrank a model with data. 7,000° is where measured data begins. A model with real data but almost no downloads sits here; everything that follows is measurement, and the room above the floor holds ten tenfold steps in downloads at 300° each. The floor is the same for every model, so it decides nothing; the degrees above it do.
7,000°
running total
2
Add the downloads.
1,236,336 download-equivalents in all, from OpenRouter tokens. Ten times the downloads adds the same 300°, so size matters but never runs away with the score.
+1,828°
8,828°
running total
3
Add the adoption rate.
About 50,937 downloads a day right now, measured between two readings, at a quarter of the weight.
+353°
9,181°
running total
=
The heat score.
Where this model stands on every list of this site. Nothing else adjusts it.
9,181°

Glossary: the terms used here

Heat score: One number for how much a model is used and wanted right now, from 0° to 9,999°. Every list on this site that says hottest is sorted by it and nothing else.
Floor: The 7,000° every model with real data starts from. The scale runs from 0° to 9,999°; the part below the floor is reserved for stand-ins (models without data), so that a model with data always ranks above one without. The floor is the same for everyone and decides nothing; the degrees above it are the measurement.
Download-equivalents: Everything the score treats as a download, added up: downloads themselves, each torrent seeder as 1,000, each AI Horde image and each run on a generation site as one.
Log scale: Counting by tens, not by ones: going from 1,000 to 10,000 downloads adds the same degrees as going from 10,000 to 100,000. It keeps giants from crowding out everything else.
Likes: Thumbs up and likes from every source, worth half a download each on the log scale.
Adoption rate: How many downloads a day a model gets right now, from the difference between two readings of a source. A quarter of the weight of the download term.
Measured and estimated: Measured means two readings of a source exist and the difference is real. Estimated means only today's total is known and the past is spread over the model's life; the site says which one it is showing.
Seeder: Someone who keeps a copy of the weights online for others to download over a torrent, or a hosted mirror doing the same. Each one counts as 1,000 downloads.
AI Horde: A volunteer network that generates images for free on donated computers. The number of images it has made with a model is public, and each counts as one download.
Generation sites: Sites that host their own copies of a model and let people generate with it, such as PixAI, SeaArt and Yodayo. Their downloads count like Civitai's and each generation run as one download.
OpenRouter tokens: A text model served through an API, such as Claude or GPT, has no weights to download, so its usage is measured in tokens: the pieces of text it read and wrote through OpenRouter, a gateway many apps use, over the last 30 days. Every 10,000,000 tokens count as one download. Source: OpenRouter (openrouter.ai/rankings), CC BY 4.0.
Full model (checkpoint): The weights themselves, the file that generates on its own. It gets a bonus because add-ons depend on it.
LoRA and embedding: Small add-on files that steer someone else's weights toward a style, a character or a concept. They cannot run alone.
Family and weights: The base a model is built on, such as SD 1.5, SDXL, Flux or Wan. Every model inherits the release date of its family's weights.
Age: Degrees taken off for how long ago the family's weights came out: 150° a year, up to 750°. It keeps old downloads that had years to pile up from freezing the top of the list.
Family boost: A small, bounded nudge for LoRAs of a family this site is pushing forward right now, always listed in the breakdown.
Stand-in score: A stable number below the floor given to a model no source reports on yet, so lists stay complete. It never ranks above a model with data.
New! (grace period): A model less than 7 days old shows New! instead of a score and holds no place in the rankings. Its score is computed from day one, but a model one day old grows faster than anything, so without the wait the week's #1 would always be whatever came out yesterday rather than what held up for a week.
Trend: How a model's growth over a period compares with the typical model's growth over the same period. Faster than typical reads as a rise, slower as a fall.
Typical model: The middle of the catalog: half the models with a figure grew faster over the period, half slower.

How this model's popularity has moved against the rest of the catalog: its growth in downloads, likes and usage over each period, measured against the typical model's growth over the same period, with its heat score at the time. Above the typical growth reads as a rise, below it as a fall. The past is reconstructed from today's totals, the model's first day and its measured download speed where two readings exist; a model that did not exist at the start of a period shows n/a.

Similar Models

Llama 3.1 8B Instruct

AI Models for Image, Video & Text Generation

Hy3 is a text model for the Text / LLM family listed on ModelTrends.ai, a read-only catalog of open source AI image, video and text models. Compare it with other Text / LLM models and check its heat score to see how it is trending.

More Text / LLM models llm tencent reasoning coding agents long context local

Browse by Model Family