New in week 43

AI inference cloud with an OpenAI-compatible API for 100+ open-source LLMs, billed per token.

Instant Linux microVMs on Deno Deploy for running untrusted code, driven from an SDK.

APIs that turn public web pages and entities into structured, verified facts for AI applications.

Platform to build, test and deploy AI agents with knowledge bases and no infrastructure to manage.

Runs AI coding agents in isolated sandboxes on a developer machine or in Docker-managed cloud, managed with the sbx CLI.

AI gateway with one OpenAI-compatible API to about 1,000 models, with per-request cost tracking and team budgets.

Build, test and deploy voice, chat and email AI agents for websites, apps and call centers, in 90+ languages.

Search API for AI agents over public web and private information, with a hosted MCP server.

Document processing infrastructure for AI agents that parses, extracts, splits and verifies documents.

Generative media model APIs and GPU compute: 1,500+ image, video, audio and 3D models behind one API.

Serverless inference API with access to 40,000+ open-weight models through one API key.

Inference and fine-tuning platform for open-source models, on serverless per-token or dedicated GPU deployments.

Launch computers for agents, code sandboxes and apps on hardware-virtualized Fly Machines.

Full Linux VMs for long-running agent tasks, with millisecond startup, live cloning and pause and resume.

Inference cloud for open-weight and custom models, with 99.99% uptime SLAs and dedicated GPU endpoints.

Managed platform on Google Cloud to build, scale, govern and optimize enterprise AI agents.

Registry of MCP servers that scans and quality-scores each one, with paid hosting on Glama.

Context layer for AI agents that ingests content and serves search, knowledge graphs and cited answers.

Fast LLM inference API that is OpenAI-compatible, for running open models.

Single OpenAI-compatible API to 100+ LLM providers, with routing, fallbacks and unified observability.

Managed Hugging Face service for deploying your own AI models to production on dedicated endpoints.

Access to hundreds of open models from partner inference providers through one Hugging Face token and SDK.

Hosting for ML demo apps built with Gradio, Docker or static HTML, with optional GPU upgrades.

AI cloud platform for training, fine-tuning and serving models on on-demand, reserved and private GPU capacity.

Cloud browsers, isolated sandboxes, web data APIs and browser agents for AI apps, accessed through APIs and SDKs.

Inference infrastructure for AI teams: hosted open-source and fine-tuned models, plus tracing and evaluation.

GPU cloud for AI training and inference, with on-demand instances, 1-Click Clusters and private clusters.

API that parses PDFs and images into grounded Markdown and extracts typed fields with a schema.

Serverless platform to build, deploy and scale AI agents with memory, RAG and hosted tools.

AI gateway and agent observability for 250+ LLMs, with tracing, analytics, guardrails and cost controls.

Framework-agnostic infrastructure for running long-running agents in production.

Web search API for AI applications, available through one API call or deployed inside your own cloud.

Cloud platform for deploying and observing voice, video and physical AI agents, built on an open-source agents framework.

Desktop app for running open LLMs locally, with a built-in agent called Bionic.

Hosted platform to deploy, observe and manage AI agents built with the open-source Mastra framework.

Managed platform for building, deploying and scaling AI agents with any framework and supported model.

Search API that makes unstructured data, from PDFs to video, searchable for agents and AI apps.

Serverless cloud for building, training and serving AI apps on GPUs, including sandboxes for agents.

Serverless sandboxes that let agents run untrusted code with fast starts, snapshots and egress controls.

Vector and hybrid search built into MongoDB, running on the same operational data as the database.

Platform for creating and managing VM-based remote runtimes for AI agents, with snapshots, SSH and a Python SDK.

OpenAI-compatible inference API for open models, with dedicated GPUs, post-training and batch inference.

Web search, extraction, crawling and research-agent APIs that return live web data to AI agents.

Runtime platform for running agents, sandboxes and production workloads on its cloud or yours.

Secure microVM and gVisor sandboxes for untrusted and agent-generated code, on Northflank Cloud or your own cloud.

Model router that predicts which LLM to call for each request, aimed at cutting coding-agent costs.

Hosted model APIs, GPU cloud and agent sandboxes on one platform, with token-based billing for models.

Containerized inference microservices for foundation models, sold as part of NVIDIA AI Enterprise and deployable on any cloud or data center.

Search, scrape, crawl and answer API for AI agents that returns clean Markdown, HTML and structured JSON.

Speech-to-speech voice agent API where the model works on audio directly, with tool calls, over WebRTC or WebSocket.