The AI cost curve didn’t just bend — it snapped. Chinese AI lab MiniMax just released M2.7, a model that scores within spitting distance of Claude Opus 4.6 and GPT-5.4 on coding benchmarks, runs on modest hardware, and costs $0.30 per million input tokens. That’s roughly 17x cheaper than Opus on input and 21x cheaper on output. But the price isn’t even the headline. The headline is how they built it: M2.7 helped build itself. ...
Remember Hunter Alpha? The mystery trillion-parameter model that appeared on OpenRouter with no name, no creator, and no explanation — sending the developer community into a frenzy of DeepSeek V4 speculation? Yeah. It was Xiaomi. A Phone Company Just Embarrassed the AI Industry On March 18, Xiaomi’s AI team MiMo confirmed that Hunter Alpha is actually an early internal test build of MiMo-V2-Pro — their agent-focused AI model. Not a chatbot. An agent brain. ...
A trillion-parameter AI model materialized on OpenRouter on March 11 with no announcement, no creator listed, and no explanation. It’s free. It has a million-token context window. And after processing 160 billion tokens in a single week, nobody can definitively say who built it. Its name is Hunter Alpha. And it might be the most fascinating AI story of 2026 so far. An AI Model That Appeared Out of Thin Air OpenRouter — the popular API gateway that routes queries across dozens of models — tagged Hunter Alpha as a “stealth model.” That’s the platform’s diplomatic way of saying: we genuinely don’t know who’s behind this. ...
Alibaba just dropped a model with 397 billion parameters that only uses 17 billion of them. And it might be the most important AI release of 2026 so far. Qwen 3.5 landed on the eve of Chinese New Year — 60% cheaper to run than its predecessor, 8x faster at decoding, and supposedly beating GPT-5.2, Claude Opus 4.5, and Gemini 3 Pro on multiple benchmarks. Bold claims. But even if you discount the scoreboard, the engineering underneath tells a story worth paying attention to. ...