Deep Learning Software
Frameworks and platforms for deep neural networks.
Platforms and libraries to train, deploy and operate models, from forecasting to text and sentiment.
What to look for
- 1Zero-shot and fine-tuning options on your own data
- 2Calibrated uncertainty, not just a point estimate
- 3Deployment on-premise or in a sovereign cloud
- 4Monitoring for drift and measurable business KPIs
Tools we connect
- OpenAI (GPT-5, GPT-4o, o-series)
Frontier language, vision and reasoning models with a mature API.
- Anthropic Claude
Safe, long-context assistant models with strong tool use and agents.
- Google Gemini & Vertex AI
Multimodal models and a managed platform to train, deploy and govern AI.
- Mistral AI
Efficient open-weight and hosted models, strong in French and multilingual tasks.
- Meta Llama
Open-weight model family for self-hosted and private deployments.
- Cohere
Enterprise models for retrieval, embeddings, classification and generation.
- AWS Bedrock
One API to many foundation models inside your AWS account.
- Azure OpenAI & Azure AI
OpenAI and Microsoft models with enterprise compliance, chatbots to vision.
- Google AI Platform
Build, deploy and manage AI models at scale on Google Cloud.
- Hugging Face
Open-source hub of models, datasets and the Transformers library.
- Replicate
Run and fine-tune open models through a simple API.
- Together.ai
Fast inference and fine-tuning for open models.