Overview
Hunyuan Hy4 Flash is the first Flash-tier model in Tencent's Hy4 generation, announced on 16 November 2026, twelve days after the full Hy4 release. It is a Mixture-of-Experts model with 224 billion total parameters, 64 billion of them Engram memory parameters, and 12 billion parameters active per token. It takes multimodal input and supports a context of up to 1 million tokens.

Tencent built Hy4 Flash for cost and speed. Pre-training combines hybrid attention, LatentMoE and Engram, which Tencent says shrinks the KV cache to 1/24 of Hy3 and 1/6 of Hy4. Post-training inherits Hy4's general ability, and reinforcement learning cuts the average output tokens per task by 50%. For serving, Tencent uses megakernel optimization, 4-bit quantization and DFly, and reports decoding above 150 tokens per second.
Tencent recommends Hy4 Flash for everyday tasks such as deep research, document processing, slide creation, simple front-end pages and computer use, and keeps the full Hy4 for complex long-horizon software engineering. On Tencent's internal WorkIndex benchmark, Hy4 Flash scores slightly above the Hy4 preview while costing 1/13 as much per task as the preview and 1/20 as much as Hy4. In a blind test with 85 WorkBuddy users across 259 tasks, Tencent says user satisfaction came close to Claude Haiku 5.5, with the lowest per-task cost and the shortest completion time of the models tested. These are company-run results.
| Released | 2026-11-16 |
|---|---|
| License | Proprietary (hosted API) |
| Weights | API only |
| Parameters | 224B total (including 64B Engram parameters), 12B active |
| Context | 1M |
| Modalities | Text, Vision |
| Status | Generally available — hosted on Tencent Cloud TokenHub and inside WorkBuddy, Yuanbao, WeCom, QQ, ima and Marvis. Tencent says OpenRouter support is coming soon and that it plans to open-source the model. |
Benchmarks

Pricing
| Input | $0.070 / 1M tokens |
|---|---|
| Cached input | $0.003 / 1M tokens |
| Output | $0.281 / 1M tokens |
Tencent's list price. Tencent Cloud TokenHub offers 50% off and WorkBuddy use is free until November 30, 2026.
Strengths
- Very low list price: $0.070 input, $0.281 output and $0.003 cached input per 1M tokens
- 1M-token context with multimodal input
- Small active size: 12B active out of 224B total parameters
- Decoding above 150 tokens per second, according to Tencent
- Tencent-reported per-task cost of 1/20 of Hy4 on its internal WorkIndex benchmark
Best for
- Deep research and document processing at low cost
- Slide (PowerPoint) generation and spreadsheet and document editing
- Simple front-end pages
- Computer-use agents for everyday office tasks
Hunyuan — every version
The full lineage of the Hunyuan line, newest first. Every version has its own page — click any to compare specs, benchmarks and pricing.
| Version | Released | Context | License |
|---|---|---|---|
| Hunyuan Hy4 Flash | 2026-11-16 | 1M | Proprietary (hosted API) |
| Hunyuan Hy4current | 2026-11-04 | 1M | Proprietary (hosted API) |
| Hunyuan Hy4 preview | 2026-08-28 | 1M | Apache-2.0 |
| Hunyuan Hy3 | 2026-07-06 | 256K | Apache-2.0 |
FAQ
Is Hunyuan Hy4 Flash open source?
Not yet. Hunyuan Hy4 Flash launched as a hosted model on Tencent Cloud TokenHub and in Tencent apps such as WorkBuddy and Yuanbao. Tencent says it plans to open-source Hy4 Flash with frequent updates, but it has not given a date or a license.
How much does the Hy4 Flash API cost?
Tencent lists Hunyuan Hy4 Flash at about $0.070 per million input tokens, $0.281 per million output tokens and $0.003 per million cached input tokens. That is roughly a twelfth of the full Hy4's $0.834 input and $2.501 output. TokenHub gives 50% off until November 30, 2026.
When should I use Hy4 Flash instead of Hy4?
Tencent recommends Hunyuan Hy4 Flash for everyday work such as deep research, document processing, slide creation, simple front-end pages and computer use. For complex, long-horizon software engineering, Tencent still recommends the full Hy4, which scores higher on its internal WorkIndex and open benchmarks.
