Thinking Machines Lab Debuts Full-Duplex 'Interaction Models' with 0.4s Response Latency | The One Dollar Digest