TechFlow News, March 14: According to a report by Whale Lab, DeepSeek V4 and a new Hunyuan model led by Tencent’s Chief AI Scientist Shunyu Yao are both expected to launch in April.
The report states that DeepSeek-V4—long refined by Wenfeng Liang as a multimodal large language model—will not only deliver a significant leap in coding capabilities but also achieve breakthroughs in LTM (Long-Term Memory).
Recently, OpenRouter launched two mysterious models: Healer Alpha and Hunter Alpha, which are speculated to be DeepSeek V4. Users with domestic IP addresses can access these two large models directly.
According to the models’ self-introductions, Hunter Alpha is a large model with 1 trillion parameters and a context window of 1 million tokens, designed specifically for agent workflows, long-horizon complex tasks, and deep tool invocation.
Healer Alpha is a fully multimodal interactive model with a context window of 260,000 tokens. It natively perceives visual and audio inputs, performs cross-modal reasoning, and executes complex multi-step tasks with precision and reliability.
As of press time, no official entity has claimed responsibility for these two large models.
Additionally, Whale Lab reports that Shunyu Yao, Tencent’s Chief AI Scientist, will also release a new Hunyuan model (at the 30-billion-parameter level) in April. This will be his first major product since joining Tencent in December last year—and is highly anticipated.
