Trace
Complete observability with environment isolation, session tracking, and
custom metadata
Recall
Scalable semantic caching for significant cost and latency savings at scale
Resilience (Coming Soon)
Automatic model fallbacks for improved reliability
Adapt (Coming Soon)
Smart model selection and cost optimization
Supported Providers
Observ works seamlessly with all major AI providers:- Anthropic - Claude models including Opus 4.5
- OpenAI - All GPT models including 5.1
- Mistral - Mistral Large and other Mistral models
- Google - Gemini models (Python SDK)
- xAI - Grok models
- OpenRouter - Access to 100+ models through a single interface
With Vercel AI SDK integration, you can access 25+ providers through a
unified API.
Two Ways to Use Observ
Vercel AI SDK
Use Vercel AI SDK’s unified API for seamless multi-provider support with
automatic provider switching
Provider SDKs
Wrap your existing provider SDK clients (Anthropic, OpenAI, etc.) directly
for drop-in observability
Key Features
Session Tracking
Group related LLM calls together using session IDs - perfect for tracking multi-turn conversations and analyzing user workflows.Learn about Sessions
Track conversation threads and multi-step workflows
Custom Metadata
Attach custom key-value pairs to traces for powerful filtering, debugging, and analytics in your dashboard.Learn about Metadata
Add context to your traces with custom metadata
Recall
Enable intelligent caching withrecall: true to automatically cache similar prompts and reduce token costs by up to 85% and achieve up to 360x latency reduction.