Freemium AI tools for AI Infrastructure & Dev Platforms

Turns documents and past sessions into a graph-plus-vector memory that AI agents can query across runs.

Runs ComfyUI workflows in a browser on rented GPUs, billing separately for edit time and generation time.

Social data API and hosted MCP server for X, Instagram, TikTok and Reddit without platform API keys

A routing and verification layer that sits between your apps and language models to cut cost and catch ungrounded answers.

EU-hosted LLM gateway with an OpenAI-compatible API and routing by price, quality or latency.

Checkpoint and LoRA hub with an on-site generator paid in Buzz

Open-source TypeScript framework for building, running, and observing AI agents in production.

Open-source search engine with vector storage and hybrid keyword-plus-semantic retrieval, self-hosted or run as a cloud service

Flat monthly plans for LLM API access with no per-token metering, built on open models.

AI gateway that routes requests to 1,600+ models through one API, with caching, fallbacks, guardrails and observability.

Postgres backend bundled with a RAG pipeline, agent runtime and a visual workflow builder

Open-source vector search engine written in Rust, available self-hosted or as a managed cloud for RAG and recommendation apps.

Open-source workspace where domain experts and engineers build test sets for AI agents together.

Tencent's model family and studio covering text, image, video and 3D, with many weights open-sourced

Open-source search and vector engine that combines text, vectors, ranking and model inference for RAG, recommendations and agents.

Connects AI agents to over 1,000 apps and tens of thousands of actions through one credential layer.

OpenAI-compatible gateway to 400+ models from one key, with routing and provider fallback.

Enterprise LLM family (Command, Embed, Rerank, Parse, Transcribe) sold API-first with VPC, on-prem and dedicated hosting options

Gives AI agents authenticated, rate-limited access to over 1,000 external tools and apps through one API.

Testing and monitoring platform for AI agents, catching hallucinations before and after they ship.

Desktop harness that joins your own devices into an encrypted mesh, keeps a private memory of your work and trains LoRA adapters on it

Simulates user scenarios against LLM agents, scores the answers and traces every step in production

A serverless platform for building, deploying, and monitoring AI agents across 600-plus LLMs.

Local context gateway that picks, filters and logs what files and tool output reach your AI coding agent or app.

Browser Studios with on-demand GPUs (T4 to H200) for training, tuning and serving PyTorch models

Runs open-weight language models locally from one command, with a hosted tier for larger models

Stores and organizes AI prompts across ChatGPT, Claude, Gemini and other models with tags, versioning and sharing.

Agent enablement platform: a scanned skill registry, security loops on every PR and evals for the coding agents your team runs

Open-source control plane that provisions and schedules GPU jobs across clouds, Kubernetes and on-prem.

Gives each AI coding agent its own persistent cloud computer — disk, shell, and network — reachable via MCP from Claude Code or Cursor.

Open-source backend platform (auth, database, storage, functions) increasingly pitched as the data layer under AI agents and RAG apps.

Documentation service and MCP server that feeds version-specific library docs and code examples into AI coding assistants.

Cloud sandboxes that give each AI agent its own isolated Linux microVM to run code, a browser or a desktop, via SDK and API.

Personal AI workspace that keeps one long-term memory of your work, people and plans and shares it across several AI models.

Hosting hub for models, datasets and demo apps

Desktop app for downloading and running open-weight LLMs locally, now with a Bionic agent and optional US-hosted cloud models.

Spreadsheet-style workbench for writing, testing and shipping prompts, with a firewall on the outputs

Tracing, analytics and prompt versioning for LLM apps, with a self-hostable community edition.

Email inboxes and delivery infrastructure for autonomous agents

Managed MongoDB cloud database with built-in vector search, hybrid search and Voyage AI embeddings for RAG and agent apps.

Source-available multimodal retrieval engine that embeds whole document pages, so RAG works over diagrams and datasheets.

Single OpenAI-compatible API and credit balance for 500+ models from 80+ providers, with fallbacks and routing rules.

Managed vector database for similarity search over billions of embeddings, with serverless and BYOC options

Simulation-driven testing, evaluation, and guardrails platform for AI agents before and after production.

Prompt versioning, evals and tracing for LLM engineering teams

Privacy-first LLM gateway that routes chat and API requests through trusted hardware.

Computer-vision platform for labeling images, training detection and segmentation models and deploying them to cloud or edge

One OpenAI-compatible API that routes requests across GPT, Claude, Gemini, and other models.

Cloud computers for running Claude Code, Codex, and other coding agents, reachable by browser, SSH, or iMessage.

Open-source Go gateway that fronts many LLM providers with one OpenAI-compatible API, fallbacks, budgets and an MCP gateway.

Enterprise control plane for AI: guardrails, observability and governance for LLM agents and ML models

Open-source machine learning plus an enterprise agentic AI platform that runs on your own hardware

Apache-licensed Python framework for building RAG pipelines and agents out of swappable components, maintained by deepset.

Open-source gateway and Python SDK that put 100+ LLM providers behind one OpenAI-style API, with keys, budgets and spend tracking.

Memory layer for AI agents and apps that stores user facts and preferences between sessions and retrieves them on demand.

A versioned filesystem built for AI coding agents, adding Git-style branching, checkpoints, and rollback to agent file storage.

French model lab offering open and commercial LLMs, the Vibe assistant, and an EU-hosted API and cloud

API of specialized models for coding agents—fast code-apply, code search and context compression—priced per million tokens.

Prebuilt GPU-optimized model containers with standard APIs for self-hosting AI models, plus hosted endpoints for prototyping.

ML platform native to Snowflake that trains, monitors, and retrains models continually instead of from scratch each cycle.