telemetry.dev is an LLM observability platform built on OpenTelemetry. It supports tracing model calls, tool steps, and retrievals, analyzing MCP traffic, and provides cost and error alerts, prompt version management, and evaluation tracing. It can be integrated via SDK or OTLP/HTTP.

Moli by Lexmount is a headless browser engine built specifically for AI agents, emphasizing fast startup and low memory usage. It can be used through cloud browser APIs, web fetch, and web search interfaces, and supports proxy and session management.

Inferock provides observable, accountable model inference through three modes—bring-your-own-key observability, managed serverless inference, and dedicated capacity. It lets you inspect response quality, latency, and usage, and offers non-cash service credits for qualifying failures.

AUDR provides an open specification along with Python/TypeScript SDKs, adapters, and sinks, enabling harnesses, routers, and billing systems to aggregate agent run usage and costs under a unified run ID, with data written to local disk first before connecting to any backend.

BytePlus is a cloud platform launched by ByteDance for overseas enterprise clients, productizing the recommendation algorithms, audio/video processing, and AI large model capabilities behind Douyin. It offers solutions such as intelligent creation, data analysis, and growth services, allowing businesses to use these battle-tested technologies on demand.

Groq is an AI inference cloud provider that develops its own LPU (Language Processing Unit) chips. Its biggest selling point is extremely low inference latency—output speeds for large models can be several to dozens of times faster than traditional GPU clouds. Developers can use the OpenAI-compatible API to access leading open-source models at blazing speed.

OpenRouter is an AI model aggregation gateway that unifies hundreds of models from providers such as OpenAI, Anthropic, Google, Meta, and Mistral into a single OpenAI-compatible API. Developers can freely switch between models, compare pricing, and benefit from automatic failover, making it a popular piece of infrastructure for building AI applications.