Anthropic says Claude Code the first Token wait time falls to 1.33 seconds
BIBIBI
AT A GLANCE
According to the report, Anthropic shortened the wait times for multiple Claude services through conversation-data compression and a pre-warming mechanism.
Article
Claude Code Response speed nearly 3 times faster: compressing conversation data and pre-warming the cloud as users type
Insight Beating AI flash news: Anthropic announced Claude its latest round of performance optimization results.
This year,August to October，Claude Code the wait time for local sessions to receive the first Token fell from 4.07 seconds to 1.33 seconds, approximately 3 times faster. One optimization uses gzip to compress the conversation data sent to the model each round, reducing network transmission overhead from repeatedly uploading long conversations.Claude Code The wait time for cloud sessions fell from 16.99 seconds to 6 seconds, approximately 2.8 times faster. In addition to compressing data, Anthropic also introduced a pre-warming mechanism. As soon as users begin typing a task, the system prepares the remote code execution environment in advance, rather than waiting to start it after the instruction is submitted.Claude The wait time for web chat and Cowork also fell from 3.36 seconds to 2.17 seconds, approximately 1.5 times faster.Claude API's first Token response time fell from 1.53 seconds to 1.01 seconds. The preceding sets of data do not include model computation time, while API data does include it.
Original link:https://m.theblockbeats.info/flash/372341
Key points
01
This year,August to October，Claude Code the wait time for local sessions to receive the first Token fell from 4.07 seconds to 1.33 seconds, approximately 3 times faster.
02
Claude Code Using gzip to compress the conversation data sent to the model each round, reducing network transmission overhead from repeatedly uploading long conversations.
03
Claude Code The wait time for cloud sessions fell from 16.99 seconds to 6 seconds, approximately 2.8 times faster.
04
When users begin typing a task, the pre-warming mechanism prepares the remote code execution environment in advance.
05
Claude The wait time for web chat and Cowork fell from 3.36 seconds to 2.17 seconds, approximately 1.5 times faster.
06
Claude API's first Token response time fell from 1.53 seconds to 1.01 seconds.
07
The data sets for local sessions, cloud sessions, and web chat and Cowork do not include model computation time, while API data includes model computation time.
AI-assisted interpretation
The following is analysis, separate from reported facts. Verify important claims independently.
These optimizations primarily reduce the wait users experience after submitting a task and before seeing the first output.gzip Compressing repeatedly sent conversation content reduces network transmission overhead; pre-warming prepares the cloud execution environment while the user is typing. Whether different metrics include model computation time is inconsistent, so all data cannot be directly compared horizontally.
Why it matters to readers
The first Token being faster means users can see the system begin responding sooner, which may particularly improve the waiting experience for long conversations and cloud code tasks.
Beginners can understand “first Token wait time” as the time between submitting a task and the interface beginning to display a response; shortening it does not mean the total task completion time is shortened by the same proportion.
Risks and unknowns
The evidence does not include Anthropic's original announcement, so the test methodology, sample size, and applicable versions cannot be verified.
The evidence does not state whether these optimizations have been rolled out to all Claude Code、Claude API and Cowork users.
The data sets other than API do not include model computation time, so changes in end-to-end task duration cannot be determined from them.
Related Developments
Loading event timeline…
Related concepts
Token
This term is not in the glossary yet. Browse related concepts in the glossary.
