Getting Started

Using the AkashChat API

The AkashChat API provides a simple interface to access the leading open-source AI models powered by the Akash Supercloud.

Akash is a permissionless marketplace for cloud resources with competitive pricing compared to traditional cloud providers.

It's easy to get started, as the AkashChat API is compatible with the OpenAI API standard. If you're already running on the OpenAI API, it's simple to make the switch. If you're just getting started, see the examples below for interacting with the API.

If you need any support, the Akash community is available around the clock to help. For immediate support, please reach out on Discord. For larger issues or feature requests, feel free to start a Discussion on GitHub.

Examples

1import openai
2import textwrap
3
4client = openai.OpenAI(
5    api_key="sk-xxxxxxxx",
6    base_url=(
7        "https://chatapi.akash.network"
8        "/api/v1"
9    )
10)
11
12response = client.chat.completions.create(
13    model="Meta-Llama-3-1-8B-Instruct-FP8",
14    messages=[
15        {
16            "role": "user",
17            "content": "Who are you?"
18        }
19    ],
20)
21
22print(textwrap.fill(
23    response.choices[0].message.content,
24    50
25))

Available Models

Chat + Completions

DeepSeek-R1
Model details
DeepSeek-R1-Distill-Llama-70B
Model details
DeepSeek-R1-Distill-Llama-8B
Model details
DeepSeek-R1-Distill-Qwen-1.5B
Model details
DeepSeek-R1-Distill-Qwen-14B
Model details
DeepSeek-R1-Distill-Qwen-32B
Model details
DeepSeek-R1-Distill-Qwen-7B
Model details
Meta-Llama-3-1-8B-Instruct-FP8
Model details
Meta-Llama-3-1-405B-Instruct-FP8
Model details
Meta-Llama-3-2-3B-Instruct
Model details
nvidia-Llama-3-1-Nemotron-70B-Instruct-HF
Model details
Meta-Llama-3-3-70B-Instruct
Model details

Embedding

BAAI-bge-large-en-v1-5
Model details