BaseModelsResponseSchema

models object[]required

List of available base model objects.

Array [

available_for_finetuningbooleanrequired

Whether this model can be used for fine-tuning jobs.

available_for_inferencebooleanrequired

Whether this model can be deployed for inference.

default_batch_sizeintegerrequired

Total batch size used during fine-tuning.

default_gradient_accumulation_stepsintegerrequired

Number of gradient accumulation steps for fine-tuning.

default_lrnumberrequired

Learning rate for fine-tuning.

default_micro_batch_sizeintegerrequired

Per-device micro-batch size for fine-tuning.

deployment_records[]required

List of active deployment records associated with this model.

display_namestringrequired

Human-readable name for the model shown in the UI.

hf_repostringrequired

Hugging Face repository identifier for this model.

model_idintegerrequired

Unique numeric identifier for the base model.

model_namestringrequired

Canonical model identifier used in API requests.

model_typestringrequired

Category of model. Common values: "language_model", "image_model".

supported_context_lenintegerrequired

Maximum context length (in tokens) supported by this model.

supported_locationsstring[]required

List of region identifiers where this model is available for deployment.

training_time_per_lognumberrequired

Estimated training time (in seconds) per log step, used for duration projections.

training_time_y_interceptnumberrequired

Y-intercept of the training time regression curve, used for fine-tuning duration estimates.

allowed_hf_lora_adapter_sizeinteger

Maximum allowed size (in bytes) for a Hugging Face LoRA adapter when fine-tuning this model.

input_token_cost_per_mnumber

Cost in USD per million input tokens for inference.

is_third_partyboolean

Whether this model was imported by the user (third-party) rather than provided natively by Hyperstack.

output_token_cost_per_mnumber

Cost in USD per million output tokens for inference.

training_cost_per_lognumber

Estimated training cost per log step, used for fine-tuning cost projections.

input_modalitiesstring[]

Input modalities the model accepts, such as text and image. A model that lists image supports image-to-text input.

output_modalitiesstring[]

Output modalities the model produces, such as text.

]

statusstring

Indicates the result of the operation. Typically "success".

Default: success

BaseModelsResponseSchema

{
  "models": [
    {
      "available_for_finetuning": true,
      "available_for_inference": true,
      "default_batch_size": "total",
      "default_gradient_accumulation_steps": "number",
      "default_lr": "learning",
      "default_micro_batch_size": "per-device",
      "deployment_records": [
        null
      ],
      "display_name": "string",
      "hf_repo": "string",
      "model_id": 0,
      "model_name": "string",
      "model_type": "string",
      "supported_context_len": 0,
      "supported_locations": [
        "string"
      ],
      "training_time_per_log": 0,
      "training_time_y_intercept": 0,
      "allowed_hf_lora_adapter_size": 0,
      "input_token_cost_per_m": 0,
      "is_third_party": true,
      "output_token_cost_per_m": 0,
      "training_cost_per_log": 0
    }
  ],
  "status": "success"
}