Mancer

About Mancer

Mancer is an uncensored large language model API for roleplay and creative writing, run as an inference service rather than a chat app. The model list names 12 hosted models, among them DeepSeek V4.1 Flash, GLM 4.7, GPT OSS 120B, Qwen3.8 27B, Magnum 72B v4 and MythoMax, with tags such as Creative, Roleplay and Coding. Context windows range from 2,560 tokens on the free MythoLite model to 1,048,576 tokens on the DeepSeek V4 family.

The service provides an OpenAI-compatible endpoint and no interface of its own: the FAQ points users to front ends such as SillyTavern, which accept Mancer as a backend. The site states there are no filters or guidelines on prompts, addressed to adult users, and its terms bar minors.

As of October 2026 an account is free and the free model needs no payment. Paid use runs on prepaid credits rather than a subscription: $4.99 buys 1,250,000 credits, $19.99 buys 5,500,000, $49.99 buys 15,000,000 and $249.99 buys 100,000,000. Each model has its own input and output cost per token; DeepSeek V4.1 Flash is listed at 0.08 credits in and 0.4 credits out. Business accounts can request weekly usage-based billing. The main limitation is the missing interface, which means a separate client app is needed, and the free model has a 150-token output cap.