Mystery model 'Hunter Alpha' on OpenRouter appears to be DeepSeek's unreleased V4 — already processing billions of tokens
A model called Hunter Alpha appeared on OpenRouter on March 11 with no developer attribution, listed as a stealth model. During Reuters tests it described itself as a Chinese AI model with training data extending to May 2025 — matching DeepSeek's existing knowledge cutoff. It advertises 1 trillion parameters and a 1 million token context window, and has already processed 160+ billion tokens since appearing.
The specs are consistent with DeepSeek V4, which the company has reportedly been preparing for an April launch. The 1M token context and trillion-parameter scale would put it squarely against GPT-5 and Claude 4 at the frontier level — and the model is currently free to access through OpenRouter. For developers building on top of API models, a DeepSeek re-entry at this capability level with free access pricing would be a direct disruption to paid API economics. Watch for an official announcement from DeepSeek.
Every story from each day, delivered at midnight UTC.