BenchChatCompletionRequest

modelModel (string)required

frequency_penalty object

anyOf

• number
• null

number

messages object[]required

Array [

roleRole (string)required

Possible values: [system, user, assistant, developer, tool, function]

content object

anyOf

• string
• ChatCompletionMessageText
• ChatCompletionMessageImageUrl
• object[]
• null

string

reasoning_content object

anyOf

• string
• null

string

name object

anyOf

• string
• null

string

tool_calls object

anyOf

• object[]
• null

Array [

idId (string)required

index object

anyOf

• integer
• null

integer

typeType (string)

Constant value: function

Default value: function

function objectrequired

idId (string)

nameName (string)required

argumentsArguments (string)required

]

tool_call_id object

anyOf

• string
• null

string

function_call object

anyOf

• object
• null

object

]

logit_bias object

anyOf

• object
• null

object

logprobs object

anyOf

• boolean
• null

boolean

top_logprobs object

anyOf

• integer
• null

integer

max_tokens object

anyOf

• integer
• null

integer

n object

anyOf

• integer
• null

integer

presence_penalty object

anyOf

• number
• null

number

response_format object

anyOf

• object
• null

object

seed object

anyOf

• integer
• null

integer

stop object

anyOf

• string
• string[]
• null

string

streamStream (boolean)

Default value: false

stream_options object

anyOf

• StreamOptions
• null

include_usageInclude Usage (boolean)

Default value: false

temperature object

anyOf

• number
• null

number

top_p object

anyOf

• number
• null

number

top_k object

anyOf

• integer
• null

integer

tools object

anyOf

• object[]
• null

Array [

property name*any

]

reasoning_effort object

anyOf

• string
• null

string

Possible values: [none, minimal, low, medium, high, xhigh]

enable_thinking object

anyOf

• boolean
• null

boolean

min_p object

anyOf

• number
• null

number

repetition_penalty object

anyOf

• number
• null

number

repetition_context_size object

anyOf

• integer
• null

integer

tool_choice object

anyOf

• string
• object
• null

string

parallel_tool_calls object

anyOf

• boolean
• null

boolean

user object

anyOf

• string
• null

string

BenchChatCompletionRequest

{
  "messages": [
    {
      "content": "Hello from Skulk",
      "role": "user"
    }
  ],
  "model": "mlx-community/Llama-3.2-1B-Instruct-4bit",
  "stream": false,
  "temperature": 0.7
}