The economical
coding agent.

Saves tokens, brain cycles, computer power, and your privacy.

╶──╮ ╶──┤ 3code v0.7.1 the economical coding agent ╶──╯ provider zaicode model glm-5.3 type a prompt. :help for commands. :q or Ctrl-D to exit. ❯  

Why 3code

3code does a few very low-hanging-fruit token usage optimizations that make a real difference, and an algorithm to keep context low

Low resource usage- more RAM, less CPU- that means longer runs on battery

A well behaved unixy interface that has just enough chrome to reflect tokens streaming but is still very quiet

Somewhat batteries included- web search is in there, sandbox is in there, cross platform

Somewhat extensible- skills and command line tools, but mostly geared to just stay out of your way

Get started

1

Get a provider

> _ deepseek
opencode deepseek free deepseek agentic coding
z.ai coding plan glm-5.3 The economical coding plan.
Used to develop 3code!
EU
tensorx glm-5.3 EU-native
use your existing subscription
supergrok grok comes with an X pro account
antigravity gemini comes with a google account
chatgpt gpt use an existing chatgpt pro account

Tip: you can use expensive models on small subscriptions with 3code. Leave the task and it will continue automatically when your 5h window resets.

3code also supports moonshot's Kimi (both subscriptions and api), Deepseek, OpenRouter, and many others.

More providers

2

Install 3code and enter your API key

$ curl -fsSL https://3code.capocasa.dev/install | sh
$ curl -fsSL https://3code.capocasa.dev/install | sh
> irm https://3code.capocasa.dev/install.ps1 | iex
$ mkdir myprojectdir
$ cd myprojectdir
$ 3code

  ╶──╮
  ╶──┤    3code v0.7.1   the economical coding agent
  ╶──╯

  no provider configured. let's add one. (ctrl+d to quit)
  supported: deepinfra, ovh, nvidia, nebius, fireworks

  api key              : ************************
  provider name or url : nvidia
  verifying... ok
  saved to ~/.config/3code/config
3

Run your first prompt

❯ Build me a Hello World program in Nim

The 3E in 3code

Carlo Capocasa

3code exists, and was made the way it is, because of a few very simple observations about AI coding- made from far away Munich, Germany, where most hype is quickly drowned in a perfectly tapped quart mug of Lager.

  1. AI programming leads to a real dilemma- you can't really afford it (it's subsidized from very high marketing budgets), but you can't afford to do without it.
  2. The subsidy is a real problem because it leads to unpredictability. Growth and enshittification- never questioned in the valley- lead to real distrust of technology as such in the rest of the world. AI is just an extreme example.
  3. The solution is efficiency and bargaining- you need to make every token count, and that's something the main AI companies just are not doing- they are optimizing for maximum performance to be impressive.
  4. The main AI companies are completely unable to shift towards efficiency in any meaningful way- it's company DNA.

The purpose of 3code is to find and use simple, common sense measures to be as token efficient as possible with existing providers, and make it as seamless as possible to switch providers to increase your bargaining power. Switching from Claude to Codex is still a big step- style is different, skills stop working. If you are used to GLM and the z.ai coding plan is down, you can just swap in mistral in the same session. This drives prices down.

3code also adheres to my- Carlo's- ideas about what good software is.

  • As simple as possible but not simpler- not minimalistic, economical. The perfect amount of features to be usable, at the smallest footprint possible for that.
  • Somewhat batteries included- you shouldn't have to install a web search plugin or worry about sandboxing, but you shouldn't need to spend tokens on sandboxing either.
  • As lean and portable as possible- it's a single binary.
  • Fancy unixy- supports scrollback properly and has nice exit codes and can be used non-interactively, but has two dynamic lines in interactive mode and updates the scrollback in small ways when it makes sense. Inspired by curl's progress.
  • Portable- usable on Windows natively, Linux, OSX, and Android via termux. More coming up!

I came up with this neat little naming shtick- the 3 in 3code. It stands for

  • 3rd party, isn't beholden to any provider, made to be independent (MIT open source, binary comes from source, no shenanigans possible)
  • and the 3 kinds of 3fficiency
    • token efficient
    • computer efficient
    • efficient with your brain cycles, so convenient.

Give it a spin and tell me what you think in https://community.3code.capocasa.dev!

E1
Token efficiency

Your bills are lower and your plan lasts longer. Free-tier tokens do real work, and you stop thinking twice before asking. Nothing else about your setup has to change: just swap the agent.

E2
Computer efficiency

No lap burn, longer battery life, and a machine that stays snappy even with dozens of agents running side by side.

E3
Ergonomics

No distractions, no gimmicks, and a learning curve so low you are productive in minutes. You stay focused, so more of what you build actually works.

Community

The forum is open at community.3code.capocasa.dev. Install help is available there, along with everything else: questions, frustrations, tips, or anything you made with 3code.

join the forum

Data to back it up

Testing token performance properly is really hard and expensive. I'm running preliminary tests that involve a 10-task subset of SWE-bench. While this is far from perfect, it does show an interesting ballpark comparison of different coding agents. 3code is doing pretty good! Working on a 40 test version now and working the way up. Did I note hardly anyone else is doing independent agent benchmarks?

SWE-bench Verified: tasks resolved
10-task subset, five agents on Z.ai GLM-5.3 via the same LiteLLM proxy, 600s per-task cap, vanilla configs
3code 7 / 10
pi 6 / 10
zcode 6 / 10
hermes 6 / 10
opencode 6 / 10
Total tokens
batch total across all 10 tasks
3code 4,209,360
pi 4,641,357
zcode 4,766,000
hermes 6,072,610
opencode 6,771,747

3code used 9% fewer tokens than pi and resolved one more task. Output tokens are the starkest gap: 71,751 for 3code vs 117,707 for pi — a 1.6× difference, and on a pay-per-token API that's money on every single turn.

Rerun: GLM-5.3, six agents incl Claude Code
same 10 tasks, 2026-09-05/06, rows ordered by total token use (prompt + output, cache-inclusive)
3code 5.1M · 9 / 10
pi 7.2M · 6 / 10
opencode 10.0M · 9 / 10
zcode 14.2M · 6 / 10
hermes 16.8M · 7 / 10
claude 23.1M · 8 / 10

Every agent hit 93–99% context-cache reads on the Z.ai coding plan, so spend tracks context volume: Claude Code processed 4.6× 3code's tokens to resolve one fewer task. 3code and opencode tie at 9/10, with 3code doing it at half the tokens. See the full report, grid and methodology.

Output tokens
batch total across all 10 tasks
3code 71,751
pi 117,707

Terse output is a design decision in 3code; pi's chattier reply style is a design decision in pi. On a pay-per-token API, that difference is money on every single turn.

Earlier round: GLM-5.2 vs opencode
same 10-task subset, same LiteLLM proxy
3code ~295k non-cached input
opencode ~1.45M non-cached input

On the same tasks with GLM-5.2, 3code yielded 4× savings vs opencode — 75% fewer tokens across the board — and resolved a task opencode failed. Zero eval errors, zero unresolved patches. See the blog post, full per-task data and methodology, and the GLM-5.2 round.

Technical Details

3code for companies

Carlo can help discover how to benefit from lower costs and more predictable AI assisted coding in your company.

schedule a call

Contributing

Patches welcome at github. Open an issue, send a PR, or just try the bleeding edge and report back.

Testing

Install the latest main branch build (no tagged release needed):

# macOS / Linux
$ curl -fsSL https://3code.capocasa.dev/main/install | sh
# Windows (PowerShell)
> irm https://3code.capocasa.dev/main/install.ps1 | iex
Founder led, founder held