Back to rankings

Claude Code Speeds Up Rendering in Long Conversations

Claude CodeView original changelog

Claude Code 2.1.257 reduced how much the terminal UI re-renders during a turn, so streaming replies no longer slow down as they grow and background-agent updates stop forcing a full-screen redraw. Prompt input got snappier too, with less rendering work happening per keystroke. A related fix stopped the token counter from freezing or crawling after switching to another subagent's transcript, and background subagents' and teammates' token counters now update live while a response streams instead of lagging behind.

Key Takeaways

  • Streaming replies no longer slow down as they grow longer, a direct result of Claude Code re-rendering less of the transcript on each update.
  • Background-agent status updates stopped forcing full-screen redraws, keeping busy multi-agent sessions from feeling sluggish elsewhere in the interface.
  • Prompt input got measurably snappier by cutting the rendering work done on every keystroke, especially noticeable in sessions with long history.
  • Token counters for background subagents and teammates now update live during streaming instead of appearing frozen.
  • Switching between subagent transcripts no longer freezes or stalls the token counter, a bug that had made usage tracking unreliable mid-session.
  • The fixes target Claude Code's heaviest use case, long-running sessions with multiple parallel subagents, where rendering overhead had compounded the most.

Less Re-Rendering, Steadier Streaming

Claude Code 2.1.257 cut the amount of UI re-rendering work done per turn in long conversations. Previously, a reply's streaming speed could visibly degrade the longer it grew, since more of the transcript had to be redrawn on every update; that no longer happens. Background-agent status updates were also changed to avoid triggering a full-screen re-render, so a busy session running several background agents does not slow down the parts of the interface a developer is actively reading.

A Snappier Prompt Input

Typing in the prompt input itself got faster, with Anthropic reducing the rendering work done on every keystroke. In sessions with a long conversation history already loaded, this had been a noticeable source of input lag; the change should make typing feel closer to instantaneous regardless of how long the current session has run.

Live Token Counters for Subagents

A related bug fix addressed the token counter freezing or crawling after a developer switched to view another subagent's transcript. Background subagents' and teammates' token counters now update live while a response streams, rather than appearing frozen until the transcript view is switched away and back, giving a more accurate real-time picture of usage across a session running multiple agents in parallel.