Skip to content
TrustList
HP
AI Model

Hy4 Preview

New· 4

Tencent's open-sourced mixture-of-experts language model with a context window above one million tokens

About Hy4 Preview

Tencent released and open-sourced Hy4 Preview, a language model, on 28 August 2026. Its mixture-of-experts design totals 770B parameters, with 49B active for each token, and Tencent says the context window exceeds one million tokens.

Where it can be used

Developers can reach the model on Tencent Cloud TokenHub and on OpenRouter, and it is built into Tencent's own products as well, among them WorkBuddy, CodeBuddy, Yuanbao and ima. Tencent offered it free on WorkBuddy and CodeBuddy for its first two weeks.

Tencent's own evaluation

Tencent's headline result comes from an internal blind evaluation in which 163 experts judged 203 engineering tasks. Hy4 Preview came out on top with a mean of 2.99 on a four-point scale, ahead of two named rivals at 2.92 and 2.94. Because this cannot be independently reproduced, it is best read as a vendor evaluation and not as a benchmark score. Tencent also reports a 31.8 percent throughput improvement from inference optimisation.

Licence

Tencent describes the model as open-sourced, yet no licence is named in what was published, so none is recorded here.

TrustList summary

Hy4 Preview is a language model released and open-sourced by Tencent on 28 August 2026. It uses a mixture-of-experts design, and Tencent describes its context window as exceeding one million tokens. It is offered through Tencent Cloud TokenHub and OpenRouter and sits inside several Tencent products.

Features

  • Mixture-of-experts architecture: 770B parameters in total, of which 49B are active for each token.
  • Long context: a window Tencent says goes beyond one million tokens.
  • API access: available via Tencent Cloud TokenHub and OpenRouter.
  • Product integration: used in WorkBuddy, CodeBuddy, Yuanbao and ima.

Key facts

  • Developer: Tencent.
  • Released: 28 August 2026.
  • Vendor evaluation: Tencent's own blind test, with 163 experts across 203 engineering tasks, gave it an average of 2.99 out of 4. This is a vendor result that cannot be independently reproduced, not a benchmark score.
  • Throughput: Tencent reports a 31.8 per cent gain from inference optimisation.
  • Licence: the announcement does not name one.

Summary written by TrustList from the business's own published information, Nov 2026.

Benchmarks & AI stats

BenchmarkOfficialCommunity avg
No benchmark scores yet. Be the first to add one.

“Official” values are editor-approved and feed the ranking. “Community avg” is the mean of member submissions (shown for transparency; it never affects the ranking until an editor approves a value).

Sign in to add a benchmark score for this model.

Hy4 Preview launches

Trust Score

4/ 100
Trust Score: New

An earned signal from verification, reviews, awards, transparency and engagement - the vendor can't buy it.

Verification
0/100 · 20%
Reviews
0/100 · 30%
Awards
0/100 · 15%
Transparency
18/100 · 20%
Engagement
0/100 · 15%
Joining soon
Recommendations
Coming soon
Complaints
Coming soon

Updated 9/16/2026

What the business tells TrustList

Founded
2026 (0 years) evidence sent

Request a demo or quote from Hy4 Preview

Protected by reCAPTCHA - Google Privacy Policy and Terms apply.

By sending, you agree we may share your request and contact details with the provider once you confirm your email.

Reviews

Write the first review of Hy4 Preview

Used it? Your experience helps other buyers decide.

Write a review

Questions & answers

No questions yet. Be the first to ask about Hy4 Preview.

Hy4 Preview alternatives

Similar listings buyers compare against this one.

Perplexity's open-weight decision model that picks among given options for routing and checks in AI systems

Google's open multimodal embedding model that maps text, images, audio and video into one vector space, under Apache 2.0

Aleph Alpha's open-weight English-German mixture-of-experts model for self-hosting, released under Apache 2.0 in October 2026

Black Forest Labs' image generation and editing model with bounding-box layouts, multiple reference images and 4K output

Cantina Security's first open-weight security model, a post-trained GLM-5.3-Flash for vulnerability research and exploit checks

Microsoft AI's text-to-speech model for keeping one voice across 23 languages, with voice cloning from short audio

Open-weight translation model from Bilibili's Index team, covering 150 languages and following glossary and style instructions

Open-source 2B decision model from AWS's Strands Agents project that picks options and returns scores with a reliability figure