Hunyuan (Tencent Hy)
TencentHunyuan is Tencent's large language model line, released under the Tencent Hy name since 2026. Hy3 is a 295B-parameter mixture-of-experts model with 21B active parameters and a 256K-token context, and Hy4 preview scales the design to 770B total and 49B active parameters with a one-million-token context. Both ship as instruct checkpoints with FP8 variants, aimed at coding, agent and office work.
Scored on the Hy releases. Tencent's video and 3D models are separate entries, hunyuan-video and hunyuan-3d.
Openness
3 high confidence- weights
- open (Hy4 preview and Hy3, with FP8 variants, ungated on Hugging Face)
- data
- closed (no pretraining corpus or data listing is published)
- code
- partial (fine-tuning and reinforcement-learning guides only, no pretraining pipeline)
- license
- Apache-2.0 (OSI)
- prior-release
- distributed (Hunyuan-A13B, an earlier release, ships under Tencent's own Hunyuan-A13B license, not Apache-2.0)
Hy4 preview and Hy3 ship their weights under Apache-2.0, ungated on Hugging Face, with fine-tuning and reinforcement-learning recipes alongside. The pretraining data and the pretraining pipeline are not published, which keeps the line at open weights.
- https://huggingface.co/api/models/tencent/Hunyuan-A13B-Instruct recorded 2026-11-12
Model API for tencent/Hunyuan-A13B-Instruct: cardData.license `other`, license_name `tencent-hunyuan-a13b`, linking the Hunyuan-A13B LICENSE on GitHub.
- https://huggingface.co/api/models/tencent/Hy3 recorded 2026-11-12
Model API for tencent/Hy3: `gated: false`, cardData.license `apache-2.0`, created 2026-07-02.
- https://huggingface.co/api/models/tencent/Hy4-preview recorded 2026-11-12
Model API for tencent/Hy4-preview: `gated: false`, cardData.license `apache-2.0`, 131 safetensors files listed, created 2026-08-27.
- https://huggingface.co/tencent/Hy3/raw/main/LICENSE recorded 2026-11-12
LICENSE: "Tencent Hy3 is licensed under the Apache-2.0", followed by the Apache License 2.0 text.
- https://huggingface.co/tencent/Hy3/raw/main/README.md recorded 2026-11-12
Card: "Hy3 provides a complete model finetuning pipeline" and supports "GRPO reinforcement learning training with verl"; "Hy3 is released under the Apache License 2.0". Nothing on the card releases or lists the pretraining data.
- https://huggingface.co/tencent/Hy4-preview/raw/main/LICENSE recorded 2026-11-12
LICENSE: "Tencent Hy4 preview is licensed under the Apache-2.0", followed by the Apache License 2.0 text.
- https://huggingface.co/tencent/Hy4-preview/raw/main/README.md recorded 2026-11-12
Card: "We open-source Hy4 preview and Hy4 preview-FP8 model weights"; its sections cover deployment, finetuning and quantization, with no pretraining data or pretraining code linked.
Adoption
4 high confidenceHugging Face downloads summed across the Hy3 and Hy4 preview checkpoints and their FP8 variants, nearly all of them for Hy3.
- https://huggingface.co/api/models/tencent/Hy3 recorded 2026-11-12
"downloads":1099861 for tencent/Hy3, the trailing 30-day count the Hub API reports.
- https://huggingface.co/api/models/tencent/Hy3-FP8 recorded 2026-11-12
"downloads":58506 for tencent/Hy3-FP8, the trailing 30-day count the Hub API reports.
- https://huggingface.co/api/models/tencent/Hy4-preview recorded 2026-11-12
"downloads":11362 for tencent/Hy4-preview, the trailing 30-day count the Hub API reports.
- https://huggingface.co/api/models/tencent/Hy4-preview-FP8 recorded 2026-11-12
"downloads":19666 for tencent/Hy4-preview-FP8, the trailing 30-day count the Hub API reports.
Capability
4 low confidenceTencent's own blind evaluations put Hy4 preview level with GLM 5.3 and Kimi K3 on engineering work, but its public benchmark tables are published only as images, and the vendor calls the release an early version with known issues. Without an independent result it sits a step below GLM rather than beside it.
- https://huggingface.co/tencent/Hy3/raw/main/README.md recorded 2026-11-12
Card: Hy3 is "a 295B-parameter Mixture-of-Experts (MoE) model with 21B active parameters"; in a blind evaluation by 270 experts it reports Hy3 at 2.67 out of 4, "outperforming GLM-5.1 at 2.51/4". Benchmark charts are embedded as images.
- https://huggingface.co/tencent/Hy4-preview/raw/main/README.md recorded 2026-11-12
Card: Hy4 preview "comprises 770B total parameters, of which 49B are activated per token"; "163 internal experts rated model outputs on 203 engineering tasks. Hy4 preview came out slightly ahead of both GLM 5.3 (2.99 vs. 2.92 ...) and Kimi K3 (2.99 vs. 2.94 ...)"; and "This is an early version of Hy4 ... we are shipping with known issues".
Verified 2026-11-12