A list of messages comprising the conversation so far. The roles of the messages must be alternating between `user` and `assistant`. The last input message should have role `user`. A message with the the `system` role is optional, and must be the very first me…
model
string
default: "meta/muse-glimmer-30b"
temperature
number
The sampling temperature to use for text generation. Muse Glimmer is a reasoning model and expects to be sampled rather than decoded greedily - the recommended pair is temperature 0.95 with top_p 1.0. Greedy decoding measurably degrades it.
0 ≤ n ≤ 1 · default: 0.95
top_p
number
The top-p sampling mass used for text generation. The top-p value determines the probability mass that is sampled at sampling time. For example, if top_p = 0.2, only the most likely tokens (summing to 0.2 cumulative probability) will be sampled. It is not reco…
The maximum number of tokens to generate in any given call. Note that the model is not aware of this value, and generation will simply stop at the number of tokens specified. Reasoning shares this budget, so leave headroom on reasoning-heavy prompts.
1 ≤ n ≤ 131072 · default: 2048
seed
integer
If specified, our system will make a best effort to sample deterministically, such that repeated requests with the same seed and parameters should return the same result.
0 ≤ n ≤ 18446744073709552000 · nullable
stream
boolean
If set, partial message deltas will be sent. Tokens will be sent as data-only server-sent events (SSE) as they become available (JSON responses are prefixed by `data: `), with the stream terminated by a `data: [DONE]` message.
Controls which (if any) tool the model calls. `none` disables tool calling; `auto` lets the model choose.
nullable · default: "auto"
stop
string[] | string
A string or list of strings where the API stops generating further tokens. The returned text does not contain the stop sequence.
nullable
frequency_penalty
number
Penalises new tokens by their existing frequency in the text so far, reducing verbatim repetition.
-2 ≤ n ≤ 2 · default: 0
presence_penalty
number
Positive values penalise tokens that have already appeared, encouraging the model to move on to new topics.
-2 ≤ n ≤ 2 · default: 0
reasoning_effort
enum
How much reasoning the model should do before answering. Higher values generally produce a longer trace in `reasoning_content`, which shares the `max_tokens` budget with the answer.
none | minimal | low | medium | high | max · default: "high"
chat_template_kwargs
object
Extra keyword arguments forwarded to the model's chat template, e.g. `{"reasoning_strength": "low"}`.
The contents of the message. <br>To pass media (only with role=`user`): <br> - Use content parts. Text is passed with `type`=`text`. <br> - Images are passed with `type`=`image_url`; set `image_url.url` to an image URL or a base64 data URI like `data:image/{fo…
The model's chain-of-thought, returned on its own channel rather than mixed into `content` (non-OpenAI extension). Streamed as `delta.reasoning_content`.