TheDrummer: Cydonia 24B V4.1

TheDrummer: Cydonia 24B V4.1 is a text model from TheDrummer, released by the provider on 2025-09-27. It is described as an uncensored and creative writing model based on Mistral Small 3.2 24B, with good recall, prompt adherence, and intelligence. It sits in the text category, so it is aimed at language tasks rather than images, audio, or other modalities.

The model takes text as input and returns text as output. It has a context window of 131072 tokens, which is large enough to hold long documents, extended conversations, or substantial reference material in a single request. The maximum output is 117964 tokens, so long-form generations are possible within that window. Its knowledge cutoff is 2024-04-30, meaning its built-in knowledge does not extend past that date.

Because it is built for creative writing and uncensored output, the practical uses follow from that focus. It suits fiction and narrative drafting, character dialogue, worldbuilding, and other open-ended prose where a permissive model is useful. Its prompt adherence and recall also make it a reasonable fit for tasks that require following detailed instructions over long inputs, such as rewriting, style matching, or working through a lengthy brief. The large context window supports multi-part projects where earlier material needs to stay in view.

You can call it through api14's OpenAI-compatible API using the model id "thedrummer/cydonia-24b-v4.1". The supported request parameters are frequency_penalty, logit_bias, logprobs, max_tokens, presence_penalty, repetition_penalty, response_format, seed, stop, structured_outputs, temperature, top_k, top_logprobs, and top_p. These cover sampling controls, penalties, deterministic seeding, stop sequences, and structured output, so you can tune generation style and constrain responses where needed.

Keep the knowledge cutoff in mind: for questions about events after 2024-04-30, the model will not have reliable built-in knowledge, and you would need to supply that information in the prompt. If your task depends on very recent facts or on modalities other than text, a different model may fit better. For text generation within its context and output limits, it is a straightforward option to test against your own prompts.