Give OpenClaw an uncensored model and it will read files, run shell commands and call tools on it, billed per token under the Unrestrictable name. The whole setup is one provider block in ~/.openclaw/openclaw.json: a base URL, an adapter name, a key variable and the model ids. Those ids also drop straight into OpenCode and Hermes Agent.
What you need
- OpenClaw installed, with its config at
~/.openclaw/openclaw.json. The file is JSON5, so comments and trailing commas are allowed, and the Gateway picks up most edits the moment you save. - An API key from the console, made for OpenClaw alone so its traffic has its own rows on the usage page.
Three steps to a first reply
- Put the key where OpenClaw will find the variable: in the process environment, or in the global env file shown here. A
.envinside your project folder is not read for provider keys.
# ~/.openclaw/.envUNRESTRICTABLE_API_KEY=sk-unr-live-…- Paste this block into
openclaw.json. ThecontextWindow,maxTokensandinputon each row come from the catalog wherever it publishes them. To add another id from the models page, append a row naming itsidand whatever else you know:
// ~/.openclaw/openclaw.json{ agents: { defaults: { model: { primary: "unrestrictable/glm-5.3-uncensored" } }, }, models: { mode: "merge", providers: { unrestrictable: { baseUrl: "https://api.unrestrictable.ai/v1", apiKey: "${UNRESTRICTABLE_API_KEY}", api: "openai-completions", models: [ { id: "glm-5.3-uncensored", name: "GLM-5.3 Uncensored" }, { id: "glm-5.3-flash-uncensored", name: "GLM-5.3 Flash Uncensored" }, { id: "deepseek-v4.1-flash-abliterated", name: "DeepSeek V4.1 Flash Abliterated" }, { id: "deepseek-v4-pro-uncensored", name: "DeepSeek V4 Pro Uncensored" }, { id: "glm-5.3-abliterated", name: "GLM-5.3 Abliterated" }, { id: "kimi-k3.1-pro-uncensored", name: "Kimi K3.1 Pro Uncensored", contextWindow: 1048576, maxTokens: 131072 }, ], }, }, },}- Confirm OpenClaw sees the provider, then chat:
openclaw models list --provider unrestrictableopenclaw models statusmodels list --provider unrestrictable echoes the ids from your block; models status names the primary it resolved and where its key came from. Switch defaults any time with openclaw models set unrestrictable/glm-5.3-flash-uncensored. Each request lands on the console’s usage page seconds later with its token counts and charge. To try a model before any of this, /chat runs the same catalog in the browser, a few free turns a day, nothing to install.
The fields, one by one
- baseUrl
- https://api.unrestrictable.ai/v1keep the /v1; OpenClaw appends /chat/completions itself
- apiKey
- ${UNRESTRICTABLE_API_KEY}resolved from the environment at start to the full sk-unr- key
- api
- openai-completionsthe Chat Completions adapter; every build on the models page speaks it
- models
- [{ id, name, … }]what this provider can serve; an id without a row here cannot be referenced
- mode
- mergeadds this provider beside OpenClaw's bundled ones instead of replacing them
- agents.defaults.model.primary
- unrestrictable/glm-5.3-uncensoredthe default model, written as provider/model
Which build to make primary
An agent resends the whole conversation plus tool output on every turn, so input tokens dominate the cost. Fields a row leaves blank get OpenClaw’s defaults: input: ["text"], reasoning: false, a context budget of 200,000 tokens, and OpenClaw’s own output cap. The cost column below prices a single turn: 30,000 tokens in, 1,000 out, at the catalog's rates on 2026-11-16. OpenClaw sends no prompt-cache hints to a custom provider, so budget for the full input price.
| Build | Per turn | 40 turns |
|---|---|---|
| GLM-5.3 Uncensored | $0.0093 | $0.37 |
| GLM-5.3 Flash Uncensored | $0.0035 | $0.14 |
| DeepSeek V4.1 Flash Abliterated | $0.03 | $1.09 |
| DeepSeek V4 Pro Uncensored | $0.03 | $1.24 |
| GLM-5.3 Abliterated | $0.07 | $2.79 |
| Kimi K3.1 Pro Uncensored | $0.04 | $1.63 |
| GLM-5.3 Uncensored | Uncensored | $0.282 | $0.846 |
| GLM-5.3 Flash Uncensored | Uncensored | $0.1053 | $0.3685 |
| DeepSeek V4.1 Flash Abliterated | Abliterated | $0.8488 | $1.6975 |
| DeepSeek V4 Pro Uncensored | Uncensored | $0.97 | $1.94 |
| GLM-5.3 Abliterated | Abliterated | $2.1825 | $4.365 |
| Kimi K3.1 Pro Uncensored | Uncensored | $1.164 | $5.82 |
If the agent is writing code, the uncensored coding AI guide puts the code builds side by side and prices a whole day of coding on each.
Test the key outside OpenClaw
Send one request by hand before blaming the config. -i prints the headers: one holds the request id, and on a non-streamed reply another holds the charge in USD.
curl -i https://api.unrestrictable.ai/v1/chat/completions \ -H "Authorization: Bearer $UNRESTRICTABLE_API_KEY" \ -H "Content-Type: application/json" \ -d '{"model": "glm-5.3-uncensored", "max_tokens": 16, "messages": [{"role": "user", "content": "Reply with the single word ok."}]}'What goes wrong
- The log warns that
${UNRESTRICTABLE_API_KEY}did not resolve: the variable is set nowhere OpenClaw looks (a project-folder.envdoes not count). Export it, or write it into~/.openclaw/.envand save. - A ref like
unrestrictable/glm-5.3-flash-uncensoredis reported as not offered: that id has no row in the provider’smodels[]. Add a row carrying theidat minimum. 401 invalid_api_key: a wrong or revoked key. Recopy it from the console and check which variable it landed in.402 insufficient_balance: credit short of the request's cost; add funds in the console.402 monthly_limit_reachedis the monthly spend cap you put on the key.404 model_not_found: the part of the ref after the slash is not an id on the models page.429 rate_limited: the key’s rate limit was hit;retry-afterin the response says when to try again.- The Gateway will not start after you edited the file: it no longer fits the schema. Unknown keys are rejected, so check every field name, or let
openclaw doctor --fixrepair it.
Where this comes from
Field names and defaults are from OpenClaw’s custom providers documentation, read 2026-11-06. Every price on the page is read from the live catalog when it renders.
Questions
One object at models.providers.unrestrictable in ~/.openclaw/openclaw.json: baseUrl https://api.unrestrictable.ai/v1, api "openai-completions", apiKey "${UNRESTRICTABLE_API_KEY}" and a models array. Then point agents.defaults.model.primary at unrestrictable/<model-id>.
In the environment variable that apiKey names, UNRESTRICTABLE_API_KEY here. OpenClaw checks the process environment, then ~/.openclaw/.env. A .env in the working directory is ignored for provider keys, and the key itself never belongs in openclaw.json.
Only a row in the provider's models[] registers a model, and it needs at least an id. agents.defaults.models only holds aliases and metadata. Add the row, save, and check with openclaw models list --provider unrestrictable.
All of them. Every id on the models page answers /v1/chat/completions in the OpenAI shape with the same key, so any of them can go into the models array spelled exactly as the page shows it. See the models page.
You pay per token at each model’s rate from a prepaid balance; a monthly plan is optional. The table on this page prices one agent-sized turn per build live. If a response reports cached input tokens, those are billed at the model’s cached rate where one exists.
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.