Claiming a $5 voucher required emailing for a code, which is more friction than the credit is worth, so most readers who would have used one never asked. List the codes directly in the EN and ZH guides instead, with the Atlas Cloud redemption link beside them. The section states no total. The list is the state, so refreshing a batch means editing the list rather than rewording the section. An email fallback stays for readers who arrive after every listed code is claimed. Update the root and core package READMEs (EN + ZH) so the sponsor bullet no longer says vouchers are requested by email. Refs #415
4.1 KiB
Atlas Cloud setup guide
Atlas Cloud is a full-modal AI inference platform that gives developers a single AI API to access video generation, image generation, and LLM APIs. Instead of managing multiple vendor integrations, you connect once and get unified access to 300+ curated models across all modalities.
OMA user vouchers
$5 Atlas Cloud vouchers for OMA users, first come first served. Take one code and leave the rest for others.
37699C92-8CCC-4224-963E-126CC2475C3B
E4D81992-9FB2-4823-A970-D90E58DB0377
29A2888B-F8B3-4157-A6EA-705781815EA6
BDA10EA7-1F2D-48EF-BF26-F37EEE325B1B
DCD47541-22BE-4314-BF6B-9EE031E5F421
Redeem a code on the Atlas Cloud Claim Your Model Rewards page.
If every code above has already been claimed, email jack@yuanasi.com with your GitHub username and a one-line use case. Supply is limited, so we can't promise one.
Disclosure: Sponsorship from Atlas Cloud; vouchers are limited and don't constitute paid endorsement of any model or feature.
Setup
Atlas Cloud exposes an OpenAI-compatible Chat Completions API ("a drop-in replacement for the OpenAI SDK"), so OMA connects to it through the built-in openai provider with a custom baseURL. This is the same pattern OMA uses for OpenRouter, Groq, Mistral, and the other OpenAI-compatible endpoints in Providers.
Environment variable
Create an API key from the Atlas Cloud console, then export it:
export ATLASCLOUD_API_KEY=your-api-key
ATLASCLOUD_API_KEY is the variable name Atlas Cloud uses in its own documentation. OMA's OpenAI-compatible setup reads it in your code and passes it to apiKey.
Agent config
Because the credential is not OPENAI_API_KEY, pass it explicitly via apiKey; the openai adapter otherwise falls back to OPENAI_API_KEY.
import { OpenMultiAgent, type AgentConfig } from '@open-multi-agent/core'
const agent: AgentConfig = {
name: 'analyst',
provider: 'openai',
baseURL: 'https://api.atlascloud.ai/v1',
apiKey: process.env.ATLASCLOUD_API_KEY,
model: 'deepseek-v4-flash-0731', // pick a current ID from the model library
systemPrompt: 'Analyze data and produce concise reports.',
tools: ['bash', 'file_read', 'file_write'],
}
const orchestrator = new OpenMultiAgent()
// Built-in filesystem tools default to a `<cwd>/.agent-workspace` sandbox;
// point the agent at an absolute path inside that root.
const result = await orchestrator.runAgent(
agent,
`Summarize the file ${process.cwd()}/.agent-workspace/report.csv`,
)
console.log(result.output)
Supported models
Atlas Cloud serves hundreds of models across LLM, image, and video modalities; OMA orchestrates the text LLMs. The catalog changes often, so treat Atlas Cloud's own listing as the source of truth instead of pinning versions here. Browse the model library for the current catalog and copy a model's exact ID string into the model field.
Current families include DeepSeek, Qwen (Alibaba), Kimi (MoonshotAI), GLM (Zhipu), MiniMax, Doubao (ByteDance), and Grok (xAI).
Because Atlas Cloud serves every model behind one OpenAI-compatible endpoint, a single Atlas Cloud key lets an OMA team mix model families across agents through the same provider: 'openai' + baseURL setup above, with no per-vendor wiring.
Atlas Cloud's image and video generation models are not text LLMs and are outside OMA's agent-orchestration scope.
Native adapter
There is no native Atlas Cloud adapter today, and the OpenAI-compatible setup above is the supported path. The adapter slot is open for Atlas Cloud to submit a PR following OMA's existing provider patterns; see packages/core/src/llm/minimax.ts for a reference implementation.
Disclosure
- Atlas Cloud is a paid sponsor of
open-multi-agent. Sponsorship does not affect technical decisions or model recommendations. - Vouchers are limited and do not constitute a paid endorsement of any model or feature.