Two free AI models, Hunter Alpha and Healer Alpha, appeared on OpenRouter on March 11 without any public announcement. The platform listed them only as “stealth models,” with no developer name, press release or blog post. That quiet launch quickly drew attention after Hunter Alpha processed more than 160 billion tokens in a week, leading developers to ask whether DeepSeek is testing its next flagship model in public. Reports Habr.com.
According to its OpenRouter page, Hunter Alpha has 1 trillion parameters and a context window of 1 million tokens. It is described as a model built for agent-style work such as long planning, complex reasoning and multi-step tasks. Healer Alpha is presented as a multimodal companion that can handle text, images and audio, with a 262,000-token context window.
Reuters reported that the chatbot described itself as a Chinese AI model trained mainly in Chinese, with a knowledge cutoff in May 2025. Some researchers say its reasoning style and system prompts, including wording about strictly following Chinese law, resemble Chinese-developed models and fit expectations for DeepSeek V4, which some local media expect in April.
Others remain unconvinced. Independent testers say the model’s token behavior, censorship patterns and weaker math performance do not fully match earlier DeepSeek systems. Another theory points to Zhipu AI, since the same anonymous OpenRouter provider previously released Pony Alpha, which later turned out to be an early GLM-5 model. For now, the identity behind Hunter Alpha and Healer Alpha remains unconfirmed.
Read “Zamin” on Telegram!
