Five models.
No refusals.
Code, security work, long context, cheap drafts — five unrestricted models are already in the picker of UNRESTRICT Desktop. Download the workspace, sign in, and ship, or point your own client at the same API.
A real coding IDE,
not a chat box.
Sessions, diffs, a terminal, and a browser in one window. The agent works in your repo and shows every change. The models are already connected, so there is no config file between you and the first prompt.
Sessions that read the repo
Give the agent a finish line. It reads the project, runs commands, and keeps working until the job is done or it needs you.
Every change in front of you
Diffs, files, terminal, and a browser sit beside the conversation. Review what changed before it lands.
The models are already connected
Five unrestricted models are in the picker on first launch. Sign in with your UNRESTRICT account and start.
Pick the model that fits the job
Five unrestricted models on one endpoint. No routing table, no hidden fallback — the id you send is the model that answers.
| Model | Focus | Context | Input / 1M | Output / 1M |
|---|---|---|---|---|
| Unrestrict DeepSeek 1017 | Flagship | 512K | $0.2143 | $0.8571 |
| Unrestrict Gemma 4 26B A4B Heretic | Heretic | 128K | $0.18 | $0.57 |
| Unrestrict Qwen 3.8 27B Derestricted | Derestricted | 128K | $0.15 | $0.90 |
| Unrestrict GLM 5.3 Flash Cybersecurity | Cybersecurity | 128K | $0.22 | $0.75 |
| Unrestrict Apex Flash 1 Abliterated | Abliterated | 128K | $0.22 | $0.75 |
Every model. Your client.
If you already have a stack, skip the IDE. Point the OpenAI SDK at a different base URL, choose a model id, and keep every other line of code.
# one endpoint, five models
curl https://unrestrict.me/v1/chat/completions \
-H "Authorization: Bearer $UNRESTRICT_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "Unrestrict DeepSeek 1017",
"messages": [{"role": "user", "content": "Explain a CVE chain."}],
"stream": true
'}Built to disappear into your stack
No SDK to learn. No new schema. Point the client you already use at a different base URL.
Drop-in
The OpenAI SDK works unchanged. Swap the base URL and key, keep every other line of code.
Streaming first
Tokens leave the model and hit your client. No buffering, no waiting for the last one to land.
Keys you control
One key per project. Revoke instantly, cap spend, and see every request in the dashboard.
Priced like infrastructure, not a subscription.
Per token, billed to the cent. No seats, no minimum commit, no annual contract. Top up five dollars and use it until it runs out.