Anthropic cuts latency across Claude Code and Messages API
Optimizations lower time to first token via gzip compression, typing-triggered session start, unbuffered requests, and reduced pre-model processing.
- Claude Code now gzips conversation payloads sent on each turn to accelerate data transfer to the model.
- Cloud sessions now initialize immediately when a user begins typing.
- The Messages API eliminated ingress request buffering and cut pre-model processing overhead to reduce initial response latency.
Developers using Claude Code and the Messages API receive faster response times and lower time to first token.
