--- title: "Novelcrafter custom API: connect an uncensored model" canonical: https://uncensoredllm.ai/guides/novelcrafter-custom-api generated: 2026-10-27T06:27:37.615Z source: https://uncensoredllm.ai/guides/novelcrafter-custom-api base_url: https://api.uncensoredllm.ai/v1 --- # Novelcrafter custom API for uncensored fiction Novelcrafter custom API setup for uncensored fiction: an OpenAI Compatible Connection, prose builds with 262K+ token windows and the cost per chapter. Steps checked October 24, 2026 · prices as of October 27, 2026 A Novelcrafter custom API connection lets the app draft with models it does not list by name. This guide adds one for uncensored prose builds, picked by context window so more of the manuscript fits in each prompt, and prices a chapter on each from the live catalog. The setting names come from Novelcrafter's help page, read on 24 October 2026. If builds with the refusals edited out are new to you, start with [AI with no restrictions](https://uncensoredllm.ai/blog/ai-with-no-restrictions), then come back here to connect one. ## What Novelcrafter is Novelcrafter is a novel-writing app that runs in the browser. You plan and write the manuscript there, keep notes on characters and places in its codex, and ask a language model to draft or rewrite prose from your scene beats. It does not sell model access itself. Its Help pages say it has no in-built credits system: you connect a vendor with your own key. Novelcrafter supports a short list of vendors by name and connects to the rest through one connection type, the OpenAI Compatible Connection. That is the one this guide uses. ## The Novelcrafter custom API connection The connection takes three values. The labels are the ones in Novelcrafter's [OpenAI API Compatible help page](https://www.novelcrafter.com/help/docs/ai-connections/openai-sdk-compatible), last updated 19 August 2026 and read on 24 October 2026. - Name: UncensoredLLM Any label; it names the connection in the list - Base URL: https://api.uncensoredllm.ai/v1 Ends in /v1, as the help page expects - API key: sk-ullm-live-… Created in the console, shown once The help page asks only for the base URL and the key; the models come next. It notes that compatible vendors usually have a base URL ending in `/v1`; this one does. ## Connect Novelcrafter step by step 1. Create a key in the [console](https://app.uncensoredllm.ai/console). It starts with `sk-ullm-` and is shown once. The balance is prepaid from $10; a key with nothing on it connects, but the first generation answers 402. 2. In Novelcrafter, open your **AI connection** settings. 3. At the bottom of the list, find **OpenAI Compatible Connection** and click **add**. 4. Type a **Name**, paste `https://api.uncensoredllm.ai/v1` as the base URL and paste your key. 5. Wait for the model list. In most cases Novelcrafter detects the models on its own. 6. If no models appear, switch to **Use custom model list** and enter the ids yourself, separated by commas or line breaks. Novelcrafter does not check for typos, so copy the ids rather than typing them. The four from this guide: ``` qwen3.8-27b-fable qwen3.8-27b-queen qwen3.8-27b-novelist gemma-4-31b-novelist ``` The help page carries one more warning. When a vendor's model names differ from the names in Novelcrafter's default collections, you need to set up a custom model collection. The ids here are this API's own, such as `qwen3.8-27b-fable`, so expect to set up that collection once. ## Which uncensored build to pick for long fiction A prompt for a scene can carry codex notes and earlier text along with the beat, and the context window caps how much of that the model sees in one call. The four builds below were picked on that basis: each carries the roleplay tag in the catalog and has a window of 262,144 tokens or more. | Model | Build | Input $/M | Output $/M | Context | Wire formats | | --- | --- | --- | --- | --- | --- | | [Qwen3.8 27B Fable](https://uncensoredllm.ai/models/qwen3.8-27b-fable) qwen3.8-27b-fable | Roleplay | $0.875 | $5.25 | 524K | openai | | [Qwen3.8 27B Queen](https://uncensoredllm.ai/models/qwen3.8-27b-queen) qwen3.8-27b-queen | Roleplay | $0.875 | $5.25 | 524K | openai | | [Qwen3.8 27B Novelist](https://uncensoredllm.ai/models/qwen3.8-27b-novelist) qwen3.8-27b-novelist | Roleplay | $0.35 | $2.10 | 262K | openai | | [Gemma 4 31B Novelist](https://uncensoredllm.ai/models/gemma-4-31b-novelist) gemma-4-31b-novelist | Roleplay | $0.35 | $1.575 | 262K | openai | Prices per million tokens, read from the live catalog on 2026-10-27. | Model id | Context window | Max output per call | | --- | --- | --- | | [`qwen3.8-27b-fable`](https://uncensoredllm.ai/models/qwen3.8-27b-fable) | 524,288 tokens | 32,768 tokens | | [`qwen3.8-27b-queen`](https://uncensoredllm.ai/models/qwen3.8-27b-queen) | 524,288 tokens | 32,768 tokens | | [`qwen3.8-27b-novelist`](https://uncensoredllm.ai/models/qwen3.8-27b-novelist) | 262,144 tokens | 32,768 tokens | | [`gemma-4-31b-novelist`](https://uncensoredllm.ai/models/gemma-4-31b-novelist) | 262,144 tokens | 32,768 tokens | Windows, output limits and prices read from the live catalog on 2026-10-27. Three of the four are Qwen 3.8 27B tunes; the rest of that family is on the [uncensored Qwen page](https://uncensoredllm.ai/uncensored/qwen). The fourth is a Gemma 4 31B tune, listed with its siblings on the [uncensored Gemma page](https://uncensoredllm.ai/uncensored/gemma). Try a beat on two of them and keep the voice you prefer; the catalog states windows and prices, not taste. Max output is the ceiling for one generation. A scene of 1,000 to 2,000 tokens sits well under it on every build here. ## How much of a manuscript fits in one prompt A window is counted in tokens, not words. A common rule of thumb for English prose is about three quarters of a word per token. The table turns each window into a rough word count with that ratio, so you can judge how much codex and earlier text a single call can carry. | Model id | Window | Roughly, in words | | --- | --- | --- | | `qwen3.8-27b-fable` | 524,288 tokens | about 393,000 | | `qwen3.8-27b-queen` | 524,288 tokens | about 393,000 | | `qwen3.8-27b-novelist` | 262,144 tokens | about 197,000 | | `gemma-4-31b-novelist` | 262,144 tokens | about 197,000 | Window × 0.75, rounded to the nearest thousand. Real ratios vary with names, dialogue and punctuation. Fitting the whole draft is rarely the point. Every token in the prompt is billed as input on every call, so a prompt that carries the relevant codex notes and the last scene or two costs a fraction of one that carries the book. The large window matters on the calls where you want the model to see much more at once, such as a rewrite that has to stay consistent with several earlier chapters. ## One key per app, with a spend limit The same key works in every client, but a separate key for Novelcrafter is easier to manage. The console lets you create several keys, set a monthly limit on a key and see your usage. If a key ends up somewhere it should not, revoke that one key and the others keep working. A limit also caps a long drafting session. Set it to what you plan to spend on the book this month. When the key reaches it, requests answer 402 with the code `monthly_limit_reached` until you raise it, and a refused request is not charged. ## What a chapter costs The worked example: a chapter drafted in 12 generations. Each sends about 30,000 tokens of codex notes and prior text, and gets about 1,000 tokens of prose back. That is an assumption, not a measurement; your prompts depend on how much context each beat carries. | Model | One chapter | A 25-chapter draft | | --- | --- | --- | | [Qwen3.8 27B Fable](https://uncensoredllm.ai/models/qwen3.8-27b-fable) | $0.38 | $9.45 | | [Qwen3.8 27B Queen](https://uncensoredllm.ai/models/qwen3.8-27b-queen) | $0.38 | $9.45 | | [Qwen3.8 27B Novelist](https://uncensoredllm.ai/models/qwen3.8-27b-novelist) | $0.15 | $3.78 | | [Gemma 4 31B Novelist](https://uncensoredllm.ai/models/gemma-4-31b-novelist) | $0.14 | $3.62 | Computed at render from the catalog prices on 2026-10-27: generations × prompt tokens × input price, plus generations × prose tokens × output price. Input dominates again: 12 × 30,000 is 360,000 input tokens against 12,000 output tokens. A large window lets you send more of the book, but you pay for what you send on every call. Trimming which codex notes a beat includes saves more than switching models. Every per-token rate is on the [pricing page](https://uncensoredllm.ai/pricing). ## Check the key and the model ids first This request lists the model ids your key can reach. It costs nothing and takes a second. Swap in your key and run it from a terminal. ```bash curl https://api.uncensoredllm.ai/v1/models \ -H "Authorization: Bearer sk-ullm-live-…" ``` A 200 with a list of ids means the key is good, and the ids in it are the exact strings for a custom model list. A 401 means the key, and the message says which way: ``` HTTP 401 {"error":{"message":"No API key was sent.","type":"authentication_error","param":null,"code":"invalid_api_key"}} ``` ``` HTTP 401 {"error":{"message":"The API key is not valid.","type":"authentication_error","param":null,"code":"invalid_api_key"}} ``` Both bodies as the API returned them on 24 October 2026. ## Troubleshooting the connection ### 401: the key "No API key was sent." means the key field is empty. "The API key is not valid." means it is wrong, cut short or revoked. Create a fresh key in the console and paste it whole. ### 404 model_not_found: the model id An id in the custom model list is not in the catalog: a typo, a display name instead of an id, or a build that has since left. Run the request above and copy the id from its output. ### 402: the balance The prepaid balance is at zero. Top up in the console; a request refused with 402 is not charged. ### Browser errors (CORS, Firefox, Safari) Novelcrafter calls the API from your browser. The API allows calls from any origin, including the Authorization header (preflight checked on 24 October 2026), so there is no proxy to set up. Novelcrafter's help page notes that some vendors implement the standard inconsistently, which can cause problems in Firefox and Safari, and suggests a Chromium-based browser such as Chrome, Brave or Vivaldi. We have not tested this connection inside Novelcrafter in Firefox or Safari. If models fail to load there but load in Chrome, use Chrome. The [daily real-model check](https://uncensoredllm.ai/proof) sends a fixed prompt to every model each day, so the model behind an id is the one on its label. Base URL, key and model id for other clients are on the [docs page](https://uncensoredllm.ai/docs). ## Questions ### Can Novelcrafter use any OpenAI-compatible API? Its OpenAI Compatible Connection accepts any vendor that follows that standard: a base URL, usually ending in /v1, and a key. Novelcrafter’s help page also lists vendors that do not work this way from the browser. ### Does Novelcrafter charge for AI? Novelcrafter’s Help says it has no in-built credits system; you connect a vendor with your own key. Model use is billed by that vendor: here per token, prepaid from $10. See per-token prices. ### Which model should I use for dark or explicit fiction in Novelcrafter? Any id from the uncensored list works through the connection. For long manuscripts, start with a build whose context window holds your codex and recent chapters; the table above lists four with windows of 262K tokens or more. See the uncensored model list. ### Why does Novelcrafter not show my models? Either the key or the base URL is wrong, or detection did not run. Check the key with the request above, then switch to Use custom model list and paste the ids. ### Does Novelcrafter work in Firefox or Safari with this API? Novelcrafter’s help page says some vendors cause problems in Firefox and Safari and suggests a Chromium-based browser. The API allows any origin, but we have not tested Novelcrafter in those two browsers. ### Is an uncensored model the same as a jailbreak prompt? No. The builds here are abliterated or uncensored fine-tunes, so the change is in the weights, not in a prompt, and there is no jailbreak text to add to your Novelcrafter prompts. See AI with no restrictions, explained. [Get a key](https://app.uncensoredllm.ai/console) Prepaid from $10, no subscription. ## New models and price changes, once a month. New uncensored builds with their prices, price moves, and what the daily model check found.