Vendor · Language models · Image models · Video models · Specialist models
Tencent has 10 models in the directory: 5 language models, 2 image models, 2 video models and 1 specialist model.
In the directory: 5 language models, 2 image models, 2 video models and 1 specialist model. 9 of them with open weights that can run locally.
Every model there ever was. Models with a profile lead to every benchmark, price and source.
| Hy-Memory-30B-A3B | Oct 15, 2026 | Open weights | n/a | n/a | n/a | 262K | 41 |
| Hy4 preview | Aug 28, 2026 | Active | n/a | n/a | n/a | 1.02M | 38 |
| Hy3 | Jul 2, 2026 | Active | n/a | $0.14 / $0.58 | 84 | 262K | 48 |
| Hy3 preview | Apr 20, 2026 | Active | n/a | n/a | n/a | 256K | 35 |
| Hunyuan A13B Instruct | Jun 27, 2025 | Active | n/a | $0.14 / $0.57 | n/a | 262K | 35 |
| Hy Image 3.5 Preview | Sep 22, 2026 | Preview | n/a | below 1 |
| HunyuanImage 3.0 Instruct | January 2026 | Active | n/a | 44 |
| Hy-Nomas | Oct 21, 2026 | Open weights | n/a | 52 |
| HunyuanVideo 1.5 | November 2025 | Active | n/a | 30 |
| Hy-SimulMT-30B-A3B | Oct 26, 2026 | Open weights | n/a | 31 |
Specialist model · Oct 26, 2026
Hy-SimulMT continuously translates spoken conversation between Chinese and English. Tencent reports scores for FLEURS, CoVoST2, ACL6060, and RealSI-Sentence, plus a long-form evaluation on RealSI. The publisher reports about three seconds of StreamLAAL. This measures streaming translation lag, not time to first audio.
Video model · Oct 21, 2026
Hy-Nomas generates an interactive world through successive video segments. Text instructions, camera paths, and prior history guide which objects, actions, and viewpoints appear in the next segment. Its diffusion backbone has eight billion parameters. The released standard configuration renders 720 × 1280 pixels at 24 frames per second and uses eight DiT denoising steps.
Language model · Oct 15, 2026
Hy-Memory-30B-A3B is a specialized language model from Tencent Hunyuan for writing persistent memories in frameworks such as Mem0, Hy-Memory, and MemOS. Hugging Face file metadata counts 30.06 billion total parameters. The model card says approximately 3 billion are active. Its config supports exactly 262,144 context tokens.
Language model · Aug 28, 2026
Hy4 preview targets long-running productivity work. The open-weight Mixture-of-Experts checkpoint activates 49 billion of 770 billion parameters per token and extends the context to 1,048,576 tokens. This checkpoint remains an early release.
Language model · Jul 2, 2026
Hy3 stands out through an additional 3.8-billion-parameter multi-token-prediction branch. Tencent combines it with 295 billion total parameters, 21 billion active parameters, 192 experts, top-8 routing, and a 262,144-token context.
Language model · Apr 20, 2026
Hy3 preview was released as an open-weight preview for practical agent work. Tencent uses a Mixture-of-Experts model with 295 billion total parameters, 21 billion active per token, and a 262,144-token context. The model supports no_think for direct responses and high for harder reasoning tasks.
Image model · January 2026
HunyuanImage 3.0 Instruct can rewrite prompts and use CoT Think in addition to text-to-image and text-image-to-image. The instruction-tuned open-weight variant has 80B total and 13B active parameters and targets precise instructions, layout, and iterative editing.
Video model · November 2025
New in HunyuanVideo 1.5 is Tencent’s compact 8.3-billion-parameter open-weight approach to text-to-video and image-to-video generation. Base outputs are 480p or 720p; a separate video super-resolution module scales them to 1080p. Native audio is absent.
Language model · Jun 27, 2025
Hunyuan A13B Instruct separates fast and slow thinking in a sparse 80-billion-parameter model with about 13 billion active per token. Fine-grained mixture-of-experts routing and grouped-query attention keep active compute limited. For agents, the model card documents a specialized vLLM parser for tool calls and internal reasoning steps.
| Vendor | Models | Model types | Best language model | Top 10 | Open weights | New in 2026 |
|---|---|---|---|---|---|---|
| Anthropic | 33 | 33 language models | Claude Opus 5.5 (Rank 1 to 3) | 4 | 0 | 13 |
| OpenAI | 63 | 46 language models, 4 image models, 2 video models, 9 audio models, 2 specialist models | GPT-6 Astra (Rank 4 to 8) | 2 | 4 | 27 |
| 66 | 43 language models, 4 image models, 6 video models, 11 audio models, 2 specialist models | Gemini 4 Carbon (Rank 1 to 5) | 2 | 19 | 26 | |
| Meta | 32 | 29 language models, 1 image model, 1 video model, 1 specialist model | Muse Wonder Full (Rank 6 to 9) | 2 | 26 | 7 |
| DeepSeek | 23 | 23 language models | DeepSeek V4.1 Flash 1017 (Rank 13 to 20) | 0 | 23 | 9 |
| Moonshot AI | 12 | 12 language models | Kimi K3.1 Pro (Rank 10 to 13) | 0 | 10 | 7 |
| xAI | 16 | 12 language models, 2 image models, 2 video models | Grok 4.7 (Rank 18 to 24) | 0 | 1 | 12 |
| Tencent | 10 | 5 language models, 2 image models, 2 video models, 1 specialist model | not ranked yet | 0 | 9 | 8 |
By release date. Grey: retired models.
8 releases since January 2026, by release date.
Prices per 1M tokens for the current language models, according to Tencent or the provider linked in each model profile.
| Current model | Input | Output | Context |
|---|---|---|---|
| Hy3 | $0.14 | $0.58 | 262K |
| Hunyuan A13B Instruct | $0.14 | $0.57 | 262K |
The directory lists 10 Tencent models: 5 language models, 2 image models, 2 video models and 1 specialist model. It counts every model Tencent ever released, including retired ones. 9 are currently available, the newest being Hy-SimulMT-30B-A3B, Hy-Nomas and Hy-Memory-30B-A3B.
Hy-SimulMT-30B-A3B, released on Oct 26, 2026. The timeline below shows every release by date.
Through the API per 1M tokens (input / output): Hy3 $0.14 / $0.58. The prices of all 2 current language models are in the price table.
Yes, 9 of the 10 models have open weights. Each model profile lists how much memory it needs.
Hy3 with 84 tokens per second according to Artificial Analysis. A typical model manages 98.
The largest is Hy4 preview with 1.02 M tokens (≈ 1,540 A4 pages). Every model's value is in the model list.
Data as of Nov 1, 2026