Skip to the stories
Wednesday, 18 November 2026
Saved

Next edition

4:02

Headlines

Industry Labs

Microsoft Research releases TokenEconomizer plugin to lower coding agent token usage

Microsoft Research announced TokenEconomizer, an open-source plugin designed to reduce token consumption and costs for AI coding agents by learning from past sessions.

FTMQ SI, written by our newsroom0 views

Share
Picture: Microsoft Research
Watch: this report in under a minute. The presenter and voice are SI-generated. More video reports

Microsoft Research reported that it has introduced TokenEconomizer, an open-source plugin created to make artificial intelligence coding agents more efficient. The plugin builds on past execution sessions to reduce operational token costs while maintaining task success rates. [1]

According to Microsoft Research, AI coding agents consume tokens during every interaction by sending instructions, tool descriptions, and session history back to their underlying models. TokenEconomizer allows these agents to spend computation more effectively rather than relearning identical lessons across multiple programming tasks. [1]

As FTMQ SI reported earlier, TokenEconomizer was developed jointly by researchers from Microsoft Research and UC Berkeley. It reduces average token costs by 18.4% to 30.9% on SWE-Bench tasks and operates alongside tools such as Claude Code, Codex, GitHub Copilot CLI, and Pi without requiring modifications to agent code or model weights. [2][5]

The release aims to address the growing compute expenses associated with autonomous software tools that help developers fix software bugs and build new features. [1]

Share

In short

  • TokenEconomizer is an open-source plugin released by Microsoft Research.
  • The tool helps AI coding agents cut token costs by building on past session history.
  • Earlier testing showed TokenEconomizer reduces average token costs by 18.4% to 30.9% on SWE-Bench tasks.

Sources

Every paragraph above points to the numbered items it rests on. Read the originals here.

  1. [1]TokenEconomizer: Learning from Experience to Make AI Coding Agents More EfficientMicrosoft Research, 1d ago (the report this story comes from)

Background

  1. [2]Claude (AI) on Wikipedia
  2. [3]Coase theorem on Wikipedia
  3. [4]Efficient-market hypothesis on Wikipedia
  4. [5]Microsoft and UC Berkeley release TokenEconomizer plugin for AI coding agents FTMQ SI, 18d ago
  5. [6]OpenAI launches ChatGPT Wallet for AI agents to make purchases FTMQ SI, 4d ago

Our newsroom writes these reports with the help of software, from the 6 sources listed and nothing else, and checks them against those sources. Facts can still be wrong or move on; the originals are the record. Spotted a mistake? Write to daniel@monsterkong.com.

Earlier reports of ours on the same people and subjects.

More in Industry Labs

Get the day in one email

Reports like this one, the top news of the last 24 hours, every morning. Free, one email a day; readers can comment under every report.

By signing up you agree to our terms and privacy policy. Unsubscribe any time.

Comments

Loading

Join the conversation

Comments are open to readers of our daily email: the top news of the last 24 hours, every morning, free. Sign up and the comment box opens.

Already on the list? Enter the same address and we will send a sign-in link.

By signing up you agree to our terms and privacy policy. Unsubscribe any time.