Counted from the engine, not the brochure
Most AI platforms give you one model. ReinaX gives you the right one per task.
These figures are counted directly from the ReinaX engine's extension registry. They are capabilities in the product, not a roadmap.
Side by side
Where a workforce platform differs from an agent product.
| Capability |  | Typical AI agent platform |
|---|---|---|
| Model providers you can route to | 33, switchable per task | Typically one, sometimes two |
| Bring your own provider keys | Rare | |
| Run fully local or self hosted models | Ollama, vLLM, SGLang | |
| Messaging and voice channels agents work in | 20 | Usually Slack and web |
| Agent count | Unlimited, you define them | Fixed catalogue or per-agent pricing |
| Agent to agent delegation | Your own org hierarchy with reporting lines | Flat or single agent |
| Builds and previews deployable software | Reina Code, live preview | |
| Video, image and motion generation | Studio | |
| Teach by demonstration | Records a browser task, saves it as a reusable skill | |
| Agent tool surface | Shell, file read and write, edit, grep, browser | Prebuilt connectors only |
| Recurring operations | Jobs, flows, cron schedules, run history | Varies |
| Pricing model | Per workspace | Commonly per seat or per agent |
The right hand column describes the common pattern across AI employee and agent builder products rather than any single vendor, because each one differs and their plans change often. Check any vendor directly before deciding.
Every provider an agent can be pointed at.
The full list of integrations, categorized for easy reference.
The full list
Every provider an agent can be pointed at.
- OpenAI
- Anthropic
- Anthropic on Vertex AI
- Google AI
- Amazon Bedrock
- Mistral AI
- DeepSeek
- Moonshot AI (Kimi)
- MiniMax
- Z.ai (GLM)
- xAI (Grok)
- Qwen
- Perplexity
- Groq
- Together AI
- OpenRouter
- Hugging Face
- NVIDIA
- Ollama
- vLLM
- SGLang
- Cloudflare AI Gateway
- Vercel AI Gateway
- BytePlus
- Volcengine
- Baidu Qianfan
- Alibaba Cloud Model Studio
- Chutes AI
- Venice AI
- Synthetic
- GitHub Copilot
- Copilot Proxy
- Kimi Coding
Channels agents work in
Integrated communication platforms.
- Telegram
- Slack
- Discord
- Microsoft Teams
- Signal
- Google Chat
- Matrix
- Mattermost
- Feishu
- LINE
- Zalo
- Nextcloud Talk
- Synology Chat
- Twitch
- IRC
- Nostr
- Tlon
- Voice calls
- Phone control
The complete ledger
Every capability, named.
Counted from the engine's own registry rather than a brochure: 5 research providers, 5 voice and speech surfaces, and ten capability groups, all live today.
Research and search
- Perplexity
- Exa
- Tavily
- Brave Search
- Firecrawl
- Browser research
- Citation-backed answers
Voice and speech
- Voice conversations
- Text to speech
- Speech to text
- Voice calls
- Phone control
Models and routing
- 33 model providers
- Per-task model switching
- Bring your own keys
- Self-hosted models
Agents and delegation
- Custom agents
- Org-chart reporting lines
- Agent-to-agent delegation
- Shared company context
Messaging channels
- Telegram
- Slack
- Discord
- Microsoft Teams
- Google Chat
- LINE
Media and documents
- Image generation
- Video generation
- Motion restyle
- Presentations
- Documents
- Spreadsheets
Developer tools
- Reina Code
- Live preview
- Shell
- File read and write
- Edit and grep
- API endpoints
Automation
- Scheduled jobs
- Flows
- Cron schedules
- Webhooks
- Run history
- Reusable skills
Data and integrations
- Browser
- Business files
- GitHub
- Supabase
- Extension registry
Governance and control
- Isolated workspaces
- Projects
- Encrypted vault
- Human approval
- Usage telemetry
Questions
Straight answers.
What is a multi-LLM AI workforce?
A workforce where each agent can be pointed at a different model, chosen per task by capability and cost, instead of every job running through one provider. ReinaX connects to 33 model providers, so a research agent can run on a reasoning model, a bulk classification agent on a cheap fast model, and a coding agent on a specialist coding model, inside the same workspace.
Which AI providers does ReinaX support?
Thirty three, including OpenAI, Anthropic, Google, Amazon Bedrock, Mistral, DeepSeek, Moonshot, MiniMax, Z.ai, xAI, Qwen, Perplexity, Groq, Together, OpenRouter and Hugging Face, plus self hosted routes through Ollama, vLLM and SGLang.
Can ReinaX run models on our own infrastructure?
Yes. Ollama, vLLM and SGLang are supported, so models can run inside your own network when data residency matters, while the rest of the platform stays managed.
How is this different from an AI employee product?
AI employee products usually sell a fixed set of prebuilt roles on one underlying model, priced per agent. ReinaX gives you the layer beneath that: you define the agents, arrange them into your own hierarchy where senior agents delegate to their reports, and choose the model per task.
Where can the agents actually work?
Twenty messaging and voice channels, including WhatsApp, Telegram, Slack, Discord, Microsoft Teams, Signal, Google Chat, Matrix and LINE, plus voice calls. Requests reach agents where your team and customers already talk.
Does more choice mean more setup?
No. Plans from Open Models upward include models with no keys to configure. Bring your own provider exists for teams that already hold contracts and want to use them.
Want a tailored AI department plan for your business?
One click — no forms. We'll email you a plan matched to your team.