Skip to main content
idapt
HomeAI ModelsPricing
Sign inStart free
All terms

Learn

What is the Aider Polyglot benchmark?

Last reviewed July 16, 2026

Aider Polyglot is a code-editing benchmark of 225 of the hardest Exercism exercises across 6 languages: C++, Go, Java, JavaScript, Python, and Rust. A model passes an exercise only when its edits are emitted in the expected format, apply cleanly to the files, and make the exercise's tests pass. It measures the practical skill coding agents live on: editing existing code reliably, not just generating snippets.

Why edit format matters

Agents change code by emitting structured edits (diffs or search-and-replace blocks) that tooling applies mechanically. A model that solves the problem but malforms its edit fails the task in practice, and this benchmark scores exactly that: correctness and format adherence together. It explains why some models that ace generation benchmarks frustrate in real editing loops.

Reading Polyglot scores

The exercises are the hardest of the Exercism sets, spread across 6 languages, so the score rewards breadth and discipline over single-language strength. Compare it with repository-scale results: SWE-bench Verified adds codebase navigation on top, while Polyglot isolates the edit mechanics.

Top models on Aider Polyglot

Full leaderboard
  1. 1GPT 588.0%
  2. 2o381.3%
  3. 3Gemini 2.5 Pro (05-06)76.9%
  4. 4DeepSeek V3.2 Exp74.2%
  5. 5o4-mini (High)72.0%
  6. 6DeepSeek R1 (0528)71.4%
  7. 7Claude Opus 470.7%
  8. 8DeepSeek V370.2%
  9. 9o161.7%
  10. 10o3-mini (High)60.4%

Peak scores from the compiled benchmark data; reasoning models use their highest effort tier.

In idapt

idapt's coding rankings include an Aider Polyglot lane, a useful signal when picking the model behind an editing agent; idapt agents apply file edits through the same structured-edit discipline the benchmark tests.

Aider Polyglot leaderboardCoding rankings

Frequently asked

Related terms

SWE-bench VerifiedLiveCodeBenchTool calling

Code Execution

Sandboxed Python, Node, and shell: agents and you, side-by-side.

CLI

Script files, agents, chats, and 200+ models: from your shell.

See it in practice

The fastest way to understand it is to use it. Start free, no credit card.

Try idapt free
  • Home
  • Pricing
  • AI Models
  • Image models
  • Voice models
  • Video models
  • Rankings
  • New models
  • Model status
  • Multi-Model Chat
  • Voice Mode
  • Voice HUD
  • Web Search
  • Image Generation
  • Video Generation
  • Audio Generation
  • Transcription
  • Drive
  • Secrets
  • Sharing
  • Workspaces
  • Tasks
  • Memory
  • Agents
  • Subagents
  • Automations
  • Skills
  • Code Execution
  • Computers
  • Computer Use
  • Cloud Computers
  • Local AI
  • AI Gateway
  • API & SDK
  • CLI
  • MCP
  • Tunnels
  • All features →
  • Blog
  • Use cases
  • Comparisons
  • Best of
  • Learn
  • Free tools
  • Prompts
  • Templates
  • Changelog
  • Help center
  • FAQ
  • Privacy
  • Compare all models
  • Support
  • Developers
  • Quickstarts
  • API reference
  • API pricing
  • CLI
  • MCP
  • Downloads
  • Desktop
  • Badges and embeds
© idapt[email protected]TermsPrivacy PolicyLegal noticeReport content
X (Twitter)