AI Infrastructure Built for Algeria
One platform for AI inference, multimodal APIs, Cloud GPU & CPU compute, developer SDKs, and production infrastructure — billed in Algerian Dinar.
Platform · Documentation · Models · Compute · Dashboard · Status
DEVUP AI is an Algerian AI infrastructure platform combining a multi-model inference gateway with on-demand Cloud GPU and CPU compute.
Developers, startups, research teams, and businesses can access leading AI models, deploy high-performance compute instances, and manage their infrastructure through one unified platform with native billing in Algerian Dinar (DZD).
DEVUP AI provides:
- OpenAI-compatible APIs
- DEVUP-native multimodal APIs
- More than 170 AI models
- On-demand Cloud GPU instances
- Scalable Cloud CPU infrastructure
- Official developer SDKs
- API key and usage management
- Real-time billing and request telemetry
- Local payment access
- Infrastructure status and operational monitoring
|
Access chat, reasoning, coding, vision, embeddings, audio, images, video, reranking, and specialized inference models through one unified API gateway. Integrate using OpenAI-compatible interfaces, the official DEVUP AI SDK, or the DEVUP Native API. |
Deploy containerized GPU and CPU instances for training, inference, fine-tuning, rendering, data processing, backend services, and general-purpose workloads. Get SSH access, JupyterLab, persistent storage options, and per-second billing. |
| Product | Description |
|---|---|
| AI Inference Gateway | One managed gateway for chat, reasoning, vision, embeddings, speech, image, video, reranking, and specialized inference. |
| OpenAI-Compatible API | Migrate compatible applications by changing the API key, base URL, and model identifier. |
| DEVUP Native API | Access model types and multimodal operations beyond standard OpenAI-compatible interfaces. |
| Cloud GPU Compute | Deploy high-performance NVIDIA GPU instances for inference, training, fine-tuning, rendering, and accelerated workloads. |
| Cloud CPU Compute | Launch scalable CPU instances for APIs, microservices, workers, data pipelines, automation, and development environments. |
| Network Volumes | Attach persistent storage that remains available across instance deployments. |
| Developer SDKs | Build with typed SDKs, API documentation, framework integrations, and production-ready examples. |
| Unified Dashboard | Manage API keys, compute instances, balance, usage, billing, and operational telemetry. |
| DZD Billing | Pay for AI APIs and compute infrastructure using supported local payment methods. |
| System Status | Monitor the operational state of the website, gateway, inference services, and documentation. |
Access more than 170 models across multiple AI modalities through one consistent infrastructure layer.
| Category | Capabilities |
|---|---|
| Language Models | Chat, reasoning, coding, structured output, tool calling, and long-context workloads |
| Vision | Image understanding, multimodal reasoning, visual question answering, and OCR |
| Embeddings | Semantic search, vector retrieval, clustering, recommendations, and RAG |
| Reranking | Improve retrieval quality by scoring and reordering search results |
| Speech Recognition | Audio transcription and multilingual speech-to-text |
| Text-to-Speech | Generate natural speech using multilingual and expressive voice models |
| Image Generation | Text-to-image generation and image editing |
| Video Generation | Text-to-video, image-to-video, and video editing |
| Specialized Inference | Classification, object detection, token classification, and model-specific operations |
- Provisioned GPU capacity
- OpenAI-compatible request formats
- Unified authentication and billing
- Model-level usage and cost tracking
- Streaming responses
- Scoped API access
- Multimodal inference
- Local DZD payments
- No model training on customer inference data
Browse the Live Model Catalog →
Launch dedicated GPU instances for:
- Large language model inference
- Model training and fine-tuning
- LoRA and adapter training
- Image and video generation
- Computer vision
- 3D rendering
- Scientific computing
- GPU-accelerated development
- Custom containers and research workloads
| GPU Class | Memory | Suitable For |
|---|---|---|
| NVIDIA B300 | 288 GB HBM3e | Frontier AI training, large-scale inference, and high-memory workloads |
| NVIDIA B200 | 180 GB | Large models, multimodal inference, training, and HPC |
| NVIDIA H200 | 141 GB | LLM inference, fine-tuning, training, and memory-intensive AI |
| NVIDIA H100 | Up to 94 GB | Enterprise AI inference, training, and accelerated computing |
| NVIDIA A100 | Up to 80 GB | Training, inference, research, and data science |
| NVIDIA L40S | 48 GB | Generative AI, rendering, inference, and visual workloads |
| RTX 6000 Ada | 48 GB | Professional AI, rendering, and workstation workloads |
| RTX 5090 | 32 GB | High-performance local inference and generative AI |
| RTX 4090 | 24 GB | Image generation, inference, development, and rendering |
Hardware availability and pricing depend on the selected region and live platform inventory.
View Available GPU Instances →
Deploy scalable CPU infrastructure for workloads that do not require dedicated GPU acceleration.
- Web applications and APIs
- Microservices
- AI agent orchestration
- Background workers
- Data processing
- Web scraping
- Automation
- CI/CD pipelines
- Databases and caching
- In-memory workloads
- Development environments
- Lightweight model serving
CPU instances provide configurable vCPU, memory, storage, and deployment options through the DEVUP AI dashboard.
|
Pay only for the compute runtime consumed by your GPU or CPU instance. |
Deploy GPU and CPU containers in seconds without long provisioning queues. |
|
Add your public SSH key and connect directly to your instance with full administrative access. |
Launch ready-to-use data science and machine learning environments with JupyterLab support. |
|
Attach Network Volumes to preserve datasets, models, checkpoints, and project files across deployments. |
Fund and manage AI and compute workloads directly in Algerian Dinar. |
Install the official SDK:
npm install devupaiCreate a client and send your first request:
import DevupAI from "devupai";
const client = new DevupAI({
apiKey: process.env.DEVUP_API_KEY!,
});
const response = await client.chat.completions.create({
model: "deepseek-ai/DeepSeek-V4-Pro",
messages: [
{
role: "system",
content: "You are a concise technical assistant.",
},
{
role: "user",
content: "Explain artificial intelligence in one paragraph.",
},
],
});
console.log(response.choices[0]?.message?.content);
console.log("Cost:", response._devup?.cost_dzd, "DZD");curl https://api.devupai.com/v1/chat/completions \
-H "Authorization: Bearer $DEVUP_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "deepseek-ai/DeepSeek-V4-Pro",
"messages": [
{
"role": "user",
"content": "Hello from Algeria!"
}
],
"max_tokens": 256
}'Configure your API key securely:
DEVUP_API_KEY=sk-devup-your-api-keyNever expose API keys in public repositories, screenshots, browser bundles, or client-side environment variables.
DEVUP AI is designed to integrate with modern AI application stacks.
| Integration | Use Case |
|---|---|
| Official Node.js SDK | TypeScript-first access to the complete DEVUP AI API |
| OpenAI SDKs | Connect existing OpenAI-compatible applications |
| Vercel AI SDK | Build streaming AI applications and Next.js interfaces |
| LangChain | Create RAG pipelines, agents, tools, and orchestration workflows |
| AutoGen | Build multi-agent applications |
| REST and cURL | Integrate from any programming language or platform |
The official SDK provides:
- Chat completions and streaming
- Image generation and editing
- Image proxy operations
- Embeddings
- Text-to-speech
- Speech transcription
- Video generation and editing
- Native model inference
- Model discovery
- Balance and health operations
- Vercel AI SDK integration
- TypeScript-first public contracts
- Zero runtime dependencies for the core REST client
View SDK Source · Install from npm
DEVUP AI is designed to process inference workloads without using customer content for model training.
- Prompts and completions are not used to train or fine-tune models.
- Inference content is processed inside isolated compute environments.
- Usage metadata such as timestamps, model, token counts, latency, and cost may be recorded for billing and platform operations.
- Usage metadata does not include prompt or completion content.
- Temporary in-memory caching may be used to improve response latency.
- Customer inference content is not written to persistent storage by the managed inference pipeline.
Read the full legal and infrastructure documentation before processing regulated or highly sensitive data.
Privacy · Security · Subprocessors · Terms
| Repository | Description |
|---|---|
devupai-node |
Official Node.js and TypeScript SDK for DEVUP AI |
.github |
Organization profile and shared GitHub configuration |
| Resource | Link |
|---|---|
| Platform | devupai.com |
| Documentation | docs.devupai.com |
| Model Catalog | devupai.com/models |
| GPU & CPU Compute | devupai.com/compute |
| GPU Instances | devupai.com/gpu-instances |
| Dashboard | devupai.com/dashboard |
| API Keys | devupai.com/dashboard/api-keys |
| System Status | status.devupai.com |
| Node.js SDK | npmjs.com/package/devupai |
| GitHub | github.com/devupai-platform |
| Support | support@devupai.com |
Build AI assistants, agents, RAG systems, search engines, voice applications, image pipelines, video products, automation platforms, research environments, and enterprise workloads on one unified infrastructure layer.
Built in Algeria. Connected to the global AI ecosystem.
Platform · Documentation · Models · Compute · Status
© 2026 DEVUP AI. All rights reserved.