Zazu · Blog
Evaluations and field notes
- Real latency on Vapi: Zazu vs the restWe put six models through the same scripted call inside identical Vapi agents, reading numbers straight off Vapi's own latency dashboard. Zazu's Qwen3.5-9B beats Vapi's fastest catalog model at a fraction of the cost.
- PhoneLLM Mini α2 on τ²-benchWhere PhoneLLM Mini α2 sits on the price–intelligence curve: a fast, low-cost peer to GPT-4.1 mini, good with tool calls, weaker at strict instruction following.