OpenRouter is the biggest list; this is the uncensored shelf of it, priced per token with nothing added at top-up and a daily check that each model is the one named. OpenRouter’s figures below are from its pricing page and FAQ as read on 2026-10-19; every figure of ours is read from the live catalog when the page renders.
The two side by side
- Model list
- OpenRouter: hundreds of models, most of them stock. Unrestrictable: 169 models, every one abliterated, uncensored, roleplay or security.
- Fee on credit
- OpenRouter: 5.5% on the Standard plan, $0.80 minimum on card, 5% on crypto (read 2026-10-19). Unrestrictable: none; top-ups of $200 and up add bonus credit.
- Prices
- OpenRouter: each provider's price passed through. Unrestrictable: a per-token price per model on the pricing page, read live.
- Is it the model named?
- OpenRouter: the FAQ states no check. Unrestrictable: a daily real-model check per model; 146 of 169 listed models read steady today.
- Wire format
- Both answer the OpenAI format at a base URL. Switching is the base URL, the key and the model id.
- Minimum
- OpenRouter: any amount plus the minimum fee. Unrestrictable: $10, credit lasts 12 months.
- Receipts
- Unrestrictable: every request on a signed receipt and a signed monthly statement you can verify without an account.
- Fallbacks and route order
- OpenRouter: provider order, fallbacks on or off, providers sorted by price, throughput or latency (read 2026-10-12). Unrestrictable: up to five backup models per request, routes sorted by price, latency or throughput, model "auto" over the uncensored tiers, and saved presets.
- Tools in the request
- Unrestrictable: web search, web fetch, remote MCP servers and a code sandbox run inside one request on 40 uncensored builds.
- Files and privacy
- Unrestrictable: PDFs and Office files readable by every listed model; secrets and personal data replaced with placeholders before the prompt leaves, per key or per request.
- Long chats
- Unrestrictable: a response cache that answers exact repeats for a tenth of the charge, and middle-out compression for chats longer than the window.
- Beyond chat
- Unrestrictable: session traces, usage CSV and log export, embeddings, speech to text, rerank and a Batch API on the same key.
OpenRouter: the baseline, list plus a fee
OpenRouter passes each provider’s price through and charges a platform fee when you buy credit: 5.5% on the Standard plan, with a $0.80 minimum on card purchases and 5% on crypto (pricing page and FAQ, read 2026-10-19). On a $100 top-up that fee is $5.50; here the same 100 dollars buys $100 of balance.
Best for: the breadth of the list and per-request provider ordering. Drawbacks for this class of model: the uncensored builds on OpenRouter are a handful among hundreds of stock models, a free tier with its own rate limits, and no stated check that the model served is the one named.
Unrestrictable: only the uncensored class, no fee on top-ups
Every one of the 169 listed models is an abliterated, uncensored, roleplay or security build; the refusal-trained models are left out. Each has a per-token price on the pricing page, read live, and nothing is added at top-up. The daily check sends a fixed prompt to every model each day and records whether the answer still matches the build. Every request lands on a signed receipt and a monthly statement you can verify.
| Wire formats | |||||
|---|---|---|---|---|---|
| GLM-5.3 Uncensored | Uncensored | $0.282 | $0.846 | — | |
| DeepSeek V4.1 Flash Abliterated | Abliterated | $0.8488 | $1.6975 | — | |
| GLM-5.3 Flash Uncensored | Uncensored | $0.1053 | $0.3685 | — | |
| DeepSeek V4 Pro Uncensored | Uncensored | $0.97 | $1.94 | — |
The same month on six builds
A busy chat client or a small app: 2,000,000 input and 500,000 output tokens in a month, at the catalog prices on 2026-11-16. The fee column is what OpenRouter adds when the same amount is bought as credit at its Standard rate; what OpenRouter’s providers charge per token for an equivalent build is on its pricing page and is not typed here.
| Model | Month here | OpenRouter fee on that amount | Daily check |
|---|---|---|---|
| GLM-5.3 Uncensored | $0.99 | $0.80 | Steady since Oct 12 |
| DeepSeek V4.1 Flash Abliterated | $2.55 | $0.80 | Steady since Oct 19 |
| GLM-5.3 Flash Uncensored | $0.39 | $0.80 | Steady since Oct 12 |
| DeepSeek V4 Pro Uncensored | $2.91 | $0.80 | Steady since Oct 12 |
How to switch from OpenRouter
Both answer the OpenAI format, so the change is the base URL, the key and the model id. Before:
client = OpenAI(base_url="https://openrouter.ai/api/v1", api_key=os.environ["OPENROUTER_API_KEY"])response = client.chat.completions.create(model="<vendor>/<model>", messages=[...])After:
client = OpenAI(base_url="https://api.unrestrictable.ai/v1", api_key=os.environ["ULLM_API_KEY"])response = client.chat.completions.create(model="glm-5.3-uncensored", messages=[...])Model ids here have no vendor prefix; the list the key can call is one free request away:
curl https://api.unrestrictable.ai/v1/models -H "Authorization: Bearer sk-unr-live-…"Create the key in the console, top up from $10, and the docs page has the same three values for SillyTavern, Janitor AI and the SDKs. Fallback lists keep the shape you know: models: [...] next to model (routing).
Where OpenRouter is still the better pick
- You need the frontier closed models and the uncensored builds on one key. This catalog lists the uncensored class only.
- You want to pick the provider by name per request. Here a request can order its routes by price, latency or throughput, and the gateway swaps a stalled one mid-request; routes are not named.
- You want a free tier. Free turns here are a few a day in the browser playground; the API is prepaid from $10.
What to check on any OpenRouter alternative
- The fee at top-up, and whether credit expires.
- Whether the price per token is listed per model or passed through from a provider you cannot see.
- Whether anything checks that the model served is the one named. A quantized or swapped build answers differently.
- What is stored: here request and response bodies are not retained by default; the usage log keeps the model, the token counts and the charge.
- Whether a failed request is charged. Here it is not.
Questions
Any OpenAI-format endpoint Janitor's proxy setting accepts. Unrestrictable lists 169 uncensored and roleplay builds priced per token with no fee at top-up; the Janitor AI guide has the three fields. See the Janitor AI guide.
Yes. Unrestrictable adds nothing at top-up; larger top-ups add bonus credit instead. Prepaid from $10, and credit lasts 12 months.
A handful, among hundreds of stock models, each at its provider’s price. Here the whole catalog is the uncensored class, and each build is checked daily.
No. Both speak the OpenAI format: change the base URL, the key and the model id. The ids here carry no vendor prefix.
After 12 months. Credit from a top-up lasts 12 months from the day it is credited.
Every model above is on the uncensored model list with its live price per million tokens. Per-token pricing sets the whole catalog side by side, and the family directory groups the builds by base model.