Aider

Cut Your Aider Token Costs by 40–70%

Terse auto-detects Aider sessions, compresses every prompt before it reaches the LLM, and monitors token usage per turn — across Claude, GPT-4o, Gemini, and local models. All on-device.

40–70%prompt compression
AutoAider detection
Any LLMmodel-aware cost tracking
0data leaves device

How Terse Works with Aider

Aider's repo-map sends thousands of tokens before you type a word. Every prompt you add goes on top. Terse compresses what it can — your typed text — and flags what it shouldn't ignore: redundant file reads as Aider applies its edits.

Terminal — Aider Session
Terse Active
$ aider --model claude/claude-sonnet-4-5
Aider v0.67.0 — Claude Sonnet 4.5 · repo-map: 8,240 tokens
● Terse detected Aider — monitoring (Sonnet 4.5 · $3.00/M input)
> can you please refactor the authetication service to use the newer asynchrnous pattern that we discussed earlier in the conversation and also make sure to add proper error handeling
Terse: 31 tokens → 14 tokens · "refactor auth service to async pattern, add error handling" · 55% reduction
Aider reading: src/auth/service.ts [2,140 tokens]
⚠ Terse: auth/service.ts read 3× this session — consider /add to keep in context
11.2Kprompt tokens
31.8Kwithout Terse
65%compressed
$0.04this turn

Auto Aider Detection

Terse scans running processes every 5 seconds. When Aider launches in your terminal, a floating monitor bar appears automatically — no setup, no configuration required.

Automatic
🗜️

Prompt Compression

Every instruction you type in Aider's terminal prompt goes through Terse's 7-stage pipeline: typo correction, filler removal, hedge stripping, and phrase shortening. Code blocks pass through unchanged.

7-Stage Pipeline
📁

Redundant File Read Flagging

Aider reads files as it applies changes — and often reads the same file multiple times per session. Terse tracks this and alerts you when a file has been read 2+ times so you can use /add to keep it in context instead.

Session Tracking
💰

Model-Aware Cost Tracking

Terse recognizes which LLM Aider is using and tracks per-turn cost at the correct model rate — Sonnet, Opus, GPT-4o, Gemini 2.5, or local models. See exactly which prompts are burning money.

Per-Turn Visibility
📊

Session Cost Summary

At the end of each Aider session, Terse shows the total token usage, total cost, and a breakdown of where tokens were spent — prompts, file reads, repo-map overhead. Know exactly what drove today's bill.

Session Analytics
🔒

100% On-Device

Terse reads your terminal input via macOS Accessibility APIs and compresses locally. No prompt text, no code, and no file contents are ever sent to Terse's servers.

Zero Cloud

Supported Aider Backends

Terse tracks token costs and monitors sessions regardless of which LLM you've configured Aider to use.

Claude Sonnet 4.5
$3/M input · $15/M output
Claude Opus 4
$15/M input · $75/M output
GPT-4o
$2.50/M input · $10/M output
Gemini 2.5 Pro
$1.25/M input · $10/M output
GPT-4o mini
$0.15/M input · $0.60/M output
Gemini 2.5 Flash
$0.075/M input · $0.30/M output
Claude Haiku 4.5
$0.80/M input · $4/M output
Local (Ollama)
Token count monitoring · no cost tracking

Frequently Asked Questions

Everything about using Terse with Aider.

Does Terse work with Aider automatically?
Yes. Terse scans running processes every 5 seconds. When Aider launches in your terminal, Terse detects it and activates a floating monitor bar. No configuration required — detection is fully automatic for all Aider startup modes.
Why is my Aider session so expensive?
The main cost drivers are: (1) Aider's repo-map — a compressed codebase summary sent at session start (5,000–50,000 tokens), (2) repeated file reads as Aider applies edits, (3) verbose prompts with filler and typos. Terse compresses your typed prompts and flags redundant file reads. The repo-map itself is not compressed — that's controlled by Aider's --map-tokens flag.
Does Terse compress Aider's repo-map?
No. Terse only compresses the prompts you type. Aider's repo-map is generated by Aider itself and sent directly to the LLM — it doesn't pass through the text fields Terse reads. To reduce repo-map tokens, use Aider's --map-tokens flag (e.g., --map-tokens 2000) or /add specific files and disable the repo-map with --no-auto-commits.
Does Terse send my Aider prompts to any server?
No. Terse is 100% on-device. It reads your terminal input via macOS Accessibility APIs, compresses locally using a Rust engine, and writes the compressed text back to the terminal input. Your prompts and code never leave your machine.

Also Works With

Terse reduces token costs across every major AI coding tool.

Claude Code ⌨️ Cursor — AI Code Editor 🤖 GitHub Copilot — VS Code 💬 ChatGPT — Browser

Stop Burning Tokens on Verbose Prompts

30-day free trial. No credit card required until your trial ends. Supports Claude, GPT-4o, Gemini, and local models.

Start Free Trial — macOS Calculate Your Savings