Platform
Models
AkashML hosts a curated set of open source AI models served via the Akash Network's decentralized GPU infrastructure. All models are accessible through the OpenAI-compatible REST API at api.akashml.com.
Available models
Models are listed in the Playground sidebar and returned by the /v1/models endpoint. Each model entry includes:
- Model ID — the identifier used in API calls (e.g.
MiniMaxAI/MiniMax-M2.5) - Context length — maximum tokens for combined prompt and completion
- Pricing — cost per million input and output tokens
Selecting a model
In the Playground, open the model selector in the sidebar to browse available models. You can filter by capability or context length.
In the API, specify the model in the request body:
{
"model": "MiniMaxAI/MiniMax-M2.5",
"messages": [{ "role": "user", "content": "Hello!" }]
}Model parameters
Each model exposes a subset of the following parameters:
| Parameter | Description |
|---|---|
temperature | Randomness of output (0 = deterministic, 2 = very random) |
max_tokens | Maximum tokens to generate in the response |
top_p | Nucleus sampling threshold |
frequency_penalty | Reduces repetition of token sequences |
presence_penalty | Encourages the model to introduce new topics |
Pricing
Credits are consumed per token. Input and output tokens are billed at different rates depending on the model. Check the current rates on akashml.com .