Overview
Tencent Hy4 is Tencent’s general-release large language model for reasoning, agentic workflows and real-world productivity. Tencent announced it on November 4, 2026. It uses a mixture-of-experts architecture with 770 billion total parameters and 49 billion active parameters. (tencent.com)
Capabilities and uses
- Accepts text and images and generates text. It supports visual understanding, complex chart and document analysis, mathematical visual reasoning, tool calling, structured output and deep reasoning.
- Supports computer-use workflows involving graphical interfaces and command-line tools. Tencent reports specialized training for professional software such as Blender and CAD.
- Intended uses include office collaboration, long-running software engineering, frontend and game development, financial analysis, cybersecurity research and scientific research.
- Tencent’s WorkBuddy evaluation reported an average score of 83.18 and a 96.15% task pass rate. These are vendor-reported results for that evaluation, not guarantees of performance in other settings. (cloud.tencent.com)
Limits and availability
Tencent Cloud TokenHub lists a 1,024K-token context window, with maximum input of 960K tokens and maximum output of 64K tokens. Its listed rate limits are 5,000,000 tokens per minute and 300 requests per minute; limits may depend on the service configuration. Outputs can still contain hallucinations or unsupported assumptions and should be reviewed, especially for consequential decisions. (cloud.tencent.com)
At announcement, Tencent said model weights and a system card would be released in the following days; the announcement alone does not establish their subsequent availability or license terms. The model is available through Tencent Cloud TokenHub and several Tencent products. (tencent.com)
API pricing
Prices below are for the Hy4 release, not Hy4 preview. They are separate regional platform prices, not currency conversions. Checked November 14, 2026.
- Tencent Cloud domestic TokenHub, Guangzhou region, online postpaid: standard pricing per 1 million tokens is CNY 6 input, CNY 18 output, and CNY 0.30 cache-hit input. The published 50% promotion applies to eligible Hy4 usage through December 31, 2026, 23:59:59 Beijing time: CNY 3 input, CNY 9 output, and CNY 0.15 cache-hit input per 1 million tokens. The domestic pricing document separately lists batch inference at CNY 3 input, CNY 9 output and CNY 0.15 cache-hit input per 1 million tokens; batch pricing is a distinct billing mode. Official domestic pricing. (cloud.tencent.com)
- Tencent Cloud International TokenHub: the official international pricing document lists standard Hy4 pricing of USD 0.834 input, USD 2.501 output, and USD 0.042 cache-hit input per 1 million tokens. Its Hy4 promotion bills eligible usage at 50% of catalog price through December 31, 2026, 23:59:59 Beijing time. Check the applicable region and final billing in the platform console. Official international pricing document. (staticintl.cloudcachetci.com)