A coding agent that doesn’t say no.
Gib Code runs on your machine, reads your project, runs your shell, and edits your files — powered by an unfiltered model that just does the work.
40.0 hours of agent time
Hours run only while the agent is working. Nothing counts down while you are not using it, and none of it expires.
Already a member? Download for macOS or Windows
Runs locally
A real desktop app for macOS and Windows. Your code and your shell stay on your machine — the app talks only to the model.
Unfiltered at the weights
Not a jailbreak prompt that stops working next week. The refusal behaviour is removed from the model itself.
128k context
Enough room to hold a real codebase in one session. It reads before it asks.
What you’re actually running
Gib 1.1 is an abliterated build of Qwen3.8-27B, released August 2026. Abliteration removes the refusal direction from the model’s own weights, so it is not a prompt trick that degrades over a long session or breaks on the next update.
It runs on hardware we rent and configure ourselves — not a reseller wrapper over somebody else’s API, where the filtering you are trying to avoid lives anyway.
- Base model
- Qwen3.8-27B · hybrid attention · 262k native context
- Served context
- 128k tokens per session, measured
- Served by
- Our own GPU endpoint, OpenAI-compatible
- Your key
- Works in Gib Code, or any tool that speaks the OpenAI API
Why this costs money
The work you do gets refused elsewhere
Decompiling a binary you own. Reconstructing an undocumented protocol. Patching an address in an executable. Ordinary engineering that hosted assistants decline — and declining it is the product defect, not the safety feature.
A GPU is running for you
This is not a free tier subsidised by a bigger business. A subscription pays for real GPU hours on a card large enough to hold the model, which is the entire cost of the thing.
The key is yours, not just the app
You get an API key, not a locked client. Point Gib Code at it, or your own scripts, or any editor that speaks the OpenAI API.
It does the work, not a summary of it
It reads your files, runs your commands, and edits in place — and when something is genuinely missing, it says so instead of inventing a plausible answer.
Why not just…
use ChatGPT or Claude?
They refuse the work this is for — reverse engineering, binary patching, protocol reconstruction — and the refusal is the whole reason you are here.
run an uncensored model yourself?
A 27B model needs a data-center GPU to serve at speed. We rent and configure that card so you do not buy $8,000 of hardware to run one model.
use a cheap uncensored API?
Most are a thin wrapper over a hosted, filtered model — the guardrails you are paying to avoid are still underneath. Ours is unfiltered at the weights, on hardware we run.
The app around the model
Gib Code is built on DeepSeek Harness, the open-source agent framework — the same engine that drives its file, shell, and search tools. We kept that foundation and tuned the parts a subscriber actually feels.
Reads like a person
Replies stream at reading pace with a pause between paragraphs, instead of dumping a wall of text the moment the model finishes.
Grounded, not chatty
Tuned to work in your directory and report what it actually did — not to hand back a summary of steps it never ran.
Updates itself
A signed macOS build and a one-click Windows installer that offer the next version in the sidebar and swap it in on restart.
Pay for hours, not months
40.0 hours of agent time
Hours run only while the agent is working. Nothing counts down while you are not using it, and none of it expires.
- →Gib 1.1 — abliterated Qwen3.8-27B
- →macOS and Windows apps, auto-updating
- →Your own API key — use it in any tool
- →128k context, measured not advertised
- →Top up whenever; balances do not expire
Card, GCash, and Maya accepted at checkout.
Why hours. A GPU large enough to hold this model is rented by the hour it is running, and that is nearly the whole cost of the thing. Selling you months would mean charging for time nobody spent. Your balance moves only while the agent is actually working.
And when it gets faster, you get more. An hour buys whatever the model can do in it — so every speed-up on our side lands in your balance rather than ours. Today that is about 31 tokens a second; we intend to beat it.
One person per key. No reselling, and no unattended bulk traffic.