In March, we focused on making production-grade LLM workflows faster, smarter, and easier to manage. We rolled out major updates to the LangDB AI Gateway: a redesigned tracing view with span-level insights, robust cost controls for scaling safely, and a smarter models interface for better deployment decisions. We also expanded our model catalog with releases like Gemini 2.5 Pro and Claude 3.7 Sonnet, and added native integrations with ClickHouse, Vercel AI SDK, and Langroid. On the community front, we connected with developers at events in Delhi and Mumbai.

🚀 LangDB AI Gateway

Tracing View Revamp

The tracing view has been fully redesigned to give you more visibility into what's happening under the hood:

Dependency Graph: Visually map the flow of your entire request, including nested calls, tool usage, and internal routing logic
Timeline View: See every span and model call across time with accurate duration and sequence
Details Panel: Click any span to explore headers, messages, tool definitions, usage & cost, outputs, and raw JSON—all in one place
Token, Cost, Duration Overview: Get high-level metrics across the top of every trace session

This makes it dramatically easier to debug multi-step agent flows, identify bottlenecks, and understand how your models are behaving in production.

Cost Control UI Enhancements

Budget management is now easier than ever with LangDB's upgraded cost control experience:

Project-Level Controls: Set daily, monthly, or total cost limits per project
Per-User & Group Limits: Restrict usage for individual users or teams (e.g., devs group) to prevent surprise bills.
Real-Time Visibility: Instantly view how much has been used, what remains, and what limits are active
No-Code Setup: Everything can be configured from the UI.

This gives teams peace of mind while scaling, and keeps infra costs transparent and under control.

Redesigned Models Section

Models Section now includes:

Filters by provider, type, input/output format, and capabilities
Visible pricing, context length, and input types
Searchable, sortable layout with inline model details

Whether you're deploying high-efficiency models like o3-mini or multi-modal powerhouses like gemini-2.5-pro, this updated section helps you pick the best tool for the job with confidence.

AI Gateway v0.2.1

Our gateway has been updated to v0.2.1 with:

Improved tracing integration
Enhanced config validation
Better support for routing across private and public LLMs

🔗 Explore on GitHub

7-Day Professional Tier Trial

You can now activate a free 7-day trial of LangDB's Professional Tier from the dashboard. Get access to full tracing, usage visibility, routing controls, and more—no commitment required.

🤖 New Models

We've expanded LangDB's model lineup to include:

Gemma 3 variants for lightweight tasks
Gemini 2.5 Pro for advanced multi-modal reasoning
Claude 3.7 Sonnet with high accuracy and long context support

All models are available instantly via LangDB's unified API

🔗 Explore All Models

🔌 Integrations

New Integrations

ClickHouse: LangDB uses ClickHouse as its data store for traces, metrics, and analytics. Native UDF support lets you run LLM calls directly from SQL. Learn about ClickHouse UDFs
Vercel AI SDK: LangDB is now available as a provider via the Vercel AI SDK. Try it on Vercel AI SDK
Langroid: Use LangDB with Langroid to build multi-agent applications with routing, memory, and modular agent logic. View examples
OpenAI Agents SDK Support: Learn how to integrate Gemini, Claude, and DeepSeek into the Agents SDK by OpenAI. Read the article

👥 Community

Delhi Meetup with ClickHouse

We co-hosted a developer meetup in Delhi alongside ClickHouse to showcase how LangDB powers scalable AI infrastructure using real-time trace analytics.

ClickHouse featured LangDB co-founder Matteo Pelati for his contributions to AI infra. Matteo presented LangDB at the ClickHouse Singapore meetup, highlighting how LangDB enables secure, observable LLM deployments while leveraging ClickHouse for storage and analytics.

🔗 See the post

Global AI Bootcamp (Mumbai)

We participated in the Global AI Bootcamp in Mumbai and hosted a talk titled "Introduction to AI Gateway", where we covered how LangDB helps teams manage multi-model routing, observability, and real-time debugging with minimal config.