Introduction

AkashML provides managed inference for open source AI models, powered by the Akash Network. You can interact with models through the Playground, or integrate via the OpenAI-compatible REST API at api.akashml.com and the Anthropic-compatible Messages endpoint at api.akashml.com/anthropic.

What you can do

• Playground — Test and compare models in a chat interface with per-model parameter controls.
• API Keys — Generate keys with configurable rate limits, concurrency caps, and expiration dates.
• Credits — Pay-as-you-go credit system with trial credits on signup, manual top-ups, and auto top-up.
• Usage Dashboard — Monitor token consumption, spending trends, and per-model cost breakdowns.
• Presets — Save model + parameter + system prompt configurations for quick reuse.

Quick start

1. Create an account at akashml.com/login and verify your email.
2. Add a payment method during onboarding to receive trial credits.
3. Try the Playground — select a model, adjust parameters, and send a message.
4. Generate an API key under Settings → API Keys to call the inference API from your code.

Your first API call

The AkashML API is OpenAI-compatible. Pass your API key in the Authorization header:

curl https://api.akashml.com/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "MiniMaxAI/MiniMax-M2.5",
    "messages": [{"role": "user", "content": "Hello!"}],
    "temperature": 0.7,
    "max_tokens": 256
  }'

Or use the OpenAI SDK with a custom baseURL:

import OpenAI from "openai"

const client = new OpenAI({
  baseURL: "https://api.akashml.com/v1",
  apiKey: "YOUR_API_KEY",
})

const response = await client.chat.completions.create({
  model: "MiniMaxAI/MiniMax-M2.5",
  messages: [{ role: "user", content: "Hello!" }],
})

console.log(response.choices[0].message.content)

Authentication

AkashML uses two authentication methods:

• Session-based — The Playground and dashboard use encrypted session cookies managed automatically when you log in via email/password or Google/GitHub OAuth.
• API key — For programmatic access. Create keys under Settings → API Keys.

API keys are passed as Bearer tokens:

Authorization: Bearer YOUR_API_KEY

Each key supports the following optional limits:

• Rate limit (RPM) — Maximum requests per minute.
• Rate limit (TPM) — Maximum tokens per minute.
• Max concurrent requests — Maximum simultaneous in-flight requests.
• Expiration date — Optional ISO 8601 date when the key stops working.

The full key value is shown only once at creation. After that, only the last 4 characters are visible — store the key somewhere safe.

Credits and billing

AkashML uses a credit-based system. Credits are consumed per token for both input and output.

• Trial credits are granted when you verify a payment method during onboarding.
• Purchase credits manually from the billing page in any amount between the configured minimum and maximum.
• Auto top-up can be enabled to automatically purchase credits when your balance falls below a threshold.
• Credit pools have optional expiration dates. Trial credits expire; purchased credits do not.

Errors and rate limiting

Errors return a JSON body with error and message fields.

When rate limited (status 429), the response includes headers:

• X-RateLimit-Limit — Total requests allowed in the window.
• X-RateLimit-Remaining — Requests remaining.
• X-RateLimit-Reset — Timestamp when the limit resets.
• Retry-After — Seconds to wait before retrying.

Rate limits are configured per API key in Settings → API Keys.

Next steps

• Browse the API Reference for full endpoint documentation.
• Open the Playground to test models, tune parameters, and save presets.
• Use Claude Code against AkashML's Anthropic-compatible Messages endpoint.