Alternatives to Groq
Groq: GroqCloud runs a short list of open models, including gpt-oss and Whisper, on Groq's own LPU hardware behind an OpenAI-compatible API, with a free tier and a pay-as-you-go Developer plan. Looking for something different? Here are other tools in Frameworks & Dev Tools.
Proa
Batteries-included Rust web framework with fast server rendering and multi-target rendering (HTML, Markdown and more) from one typed component.
Sygal
Local-first Mac and Windows app plus CLI that converts PDFs, Office files, scans and web pages into clean Markdown or JSON for AI prompts and pipelines.
Weatherdeck
Open-source, read-only status board for TV or phone showing what your AI coding agent lanes and machines are doing and what is waiting on you.
Sphinx
Free, MIT-licensed Git hook that quizzes you on what your coding agent's diff actually does before a push to protected branches goes through; works with Claude Code, Codex, Cursor or plain Git.
llmspeed
Open-source simulator that lets you feel a local LLM's tokens/sec and time-to-first-token on a chosen GPU or Mac before you buy the hardware, based on community benchmarks.
esp32sim
Open-source (MIT) ESP32 emulator written in Rust that runs real firmware locally or in the browser, with no hardware or cloud needed.
siltop
Rust terminal system monitor for Apple Silicon Macs showing CPU, GPU, power, clocks, temperatures and processes, without sudo or telemetry.
WhatsApp MCP
Open-source (MIT), read-only MCP server you deploy to your own Cloudflare account so Claude Code or any MCP client can search and read your WhatsApp chats; it can't send messages and is unofficial (uses the WhatsApp Web protocol).
OmniPTY
Open-source (MIT), GPU-rendered Rust terminal for macOS and Linux with built-in splits, pinnable workspaces, a vim-style file tree, copy mode and a git-aware status bar; no account or telemetry.
Spotnote
Open-source (MIT) Vite plugin: click any element in your running app, type what should change, and get a prompt with the exact source file, line and column to paste into Claude Code, Cursor or Codex; local-only and dev-mode only.
Death Plan Switch
Free, MIT-licensed, self-hosted dead man's switch for solo developers: it emails you weekly and, if you stop checking in, asks your next of kin and then notifies the contacts who need to take over; PHP 8.1+, no database.
Bifrost
Open-source (Apache-2.0) AI gateway from Maxim AI: one OpenAI-compatible API for thousands of models with failover, load balancing, semantic caching, virtual keys, budgets and an MCP gateway; Enterprise plan available.
Gauge CLI
Open-source (MIT) CLI that runs real coding agents against your Skills, CLIs, SDKs, MCP servers and docs, captures where they get stuck and lets your own agent test fixes locally in Docker; part of the Gauge agent-led growth platform.
e2e by TesterArmy
Open-source (Apache-2.0) end-to-end testing framework for web, iOS and Android: describe a goal in plain language, an agent drives the app, and assertions verify the result; recorded steps replay with no model calls. From YC-backed TesterArmy.
Chiplab
MCP platform from Veecle that lets AI agents compile, simulate and validate firmware on the actual target chip without physical hardware; in beta.
Opper
EU-hosted AI gateway: 700+ models behind one OpenAI-compatible API, pay-as-you-go, with spend caps, model allowlists and zero data retention; works with Claude Code, OpenCode and Cursor.
Reticle
AI app testing that runs inside your app: it walks a flow like a user and reads network calls, app state and the console so your coding agent can verify its own work; free open-source SDK for Claude Code, Cursor and any MCP agent.
Petal
Free, open-source pair programming app from the makers of CoScreen: low-latency, high-resolution window sharing with remote control, for Mac and Windows.
pocketty
SSH terminal for iPhone and iPad made for herdr: end-to-end sealed alerts when a coding agent is blocked or done, keys in the Secure Enclave, and real keystrokes to answer it. 14-day trial, then a one-time purchase.
Jotbus
End-to-end encrypted shared scratchpad that lets Claude Code, Codex, Cursor, Gemini and other coding agents share notes and files, hand off work and review changes across machines; try it with npx jotbus.
k10s
Open-source (Apache-2.0) Kubernetes terminal UI you can click: searchable, themeable, a single static binary for macOS, Linux and Windows, with an AI assistant that knows your cluster, namespace and selected object.
DocuMemory
Memory engine for AI-first engineers: distils Claude Code, ChatGPT and Cursor sessions into architecture decisions and prompt-injectable context blocks, recalled over MCP or a REST API; 15 free distillations, then from $10/month.
Etchv
Invisible forensic watermarking API for images, PDFs and video, written by a neural model and detectable after resizing and cropping; SDKs for Python, Node.js, Go, Rust, C# and Java plus an MCP server.
Staffcoder
Hands-on coding practice in a browser IDE: 1,000+ production-style challenges across 40+ learning paths, from React, Next.js and Spring Boot to DevOps and AI engineering.
Nace
Execution layer for agents: a Perception API that reads any file and a Decision API that makes small calls in under 200 ms on fast, low-cost models, under one API key; Drex models with open weights.
OpenLume
Ask a technical question and get a personalized, narrated explainer video with diagrams, code and animations that you can interrupt with follow-up questions.
ProvenMetal
Quote and order assembled PCBs from Codex or Claude through the ProvenMetal MCP server, at JLCPCB's own instant price with design checks and sourcing of parts JLCPCB doesn't stock.
Pomerado
Turns any website, from consumer sites to legacy portals, into a hosted API or MCP server for your agents, faster and more reliable than browser agents; open source and backed by Y Combinator.
NitroStack
Full-stack MCP platform: open-source TypeScript and Python SDKs, an MCP CLI, the NitroStudio desktop app for testing and debugging, and NitroCloud managed hosting for production MCP servers and apps.
Context.dev
Web data API for AI agents: scrape any URL into LLM-ready Markdown, search the live web, crawl and map sites, monitor pages for changes and pull brand data, with SDKs, a CLI and an MCP server; free credits to start.
Zernio
Marketing infrastructure for products and agents: one API and a hosted MCP server for social posts, ads, a unified inbox, WhatsApp, SMS and calls, built on official platform APIs; free up to two accounts.
Inferock
Accountable LLM inference: watch your own OpenAI, Anthropic or Gemini calls for failures and overbilling, run serverless inference with credits back on eligible failed calls, or rent dedicated capacity; open-source inferock-bench proxy.
RunInfra
Open-model APIs built for agents: fast responses, warm replicas with up to 98% cache hits for repeated context, and one API key that works with the OpenAI and Anthropic SDKs. Backed by Y Combinator.
Gleap
AI-native help desk for software teams: in-app bug reporting and session replay plus Kai agents that answer customers, investigate tickets and open code fixes your team reviews; plans from $29/month.
Beam
Puts windows from one Mac onto another as native windows with real input, so agents and builds can keep running on a Mac mini or Studio while you work from your laptop; encrypted, direct over your own network, subscription or lifetime license.
Invofox
Document parsing API that turns invoices, payslips, contracts and any document into validated JSON or Markdown for RAG and LLM pipelines, with classification, splitting, confidence scores and continuous learning; SOC 2 and ISO 27001.
Siteprint
Safari extension for Mac that measures any website's design (colours, type scale, spacing, corners and layout) and exports it as one exact prompt or DESIGN.md for Claude Code, Cursor or Codex; free to try, Pro $19.99 once.
Maximem Synap
Memory and context layer for AI agents with entity resolution, temporal reasoning and scoping built in, sub-15ms recall, a hosted MCP endpoint and native adapters for 23 agent frameworks including LangChain and the Claude Agent SDK; free tier, paid from $19/month.
Hopscotch
One API for 500+ models from Anthropic, OpenAI, Google and more, with spend limits per key, teammate and workspace, guardrails, fallbacks, request logs and bring-your-own-key; pay as you go at list price.
iFixAi
Independent auditing for AI agents: finds where an agent hides information, overrides decisions or takes unauthorised actions behind a successful result, with recurring audits and reports per agent.
Coddy
Learn to code with bite-sized interactive lessons and playgrounds for Python, JavaScript, SQL and 15+ languages, plus learning paths for software, web and game development; 5M+ learners, optional premium plan.
NOAN
Fact infrastructure for agentic companies: one API over REST and MCP for your verified business facts, read live by every agent, app and site; free plan, paid from $99/month.
Karada.ai
Infrastructure for MCP: auto-compile APIs into Go MCP servers, keep tools in sync with CI/CD on every deploy, add composable plugins and multiplex everything behind one gateway; free Hobby plan, Pro from $19/month.
CoIsland
Mac app that watches your engineering stack (GitHub Actions, Sentry and more), shows only what needs you and hands the work to Claude Code, Codex or Cursor; keys stay on your Mac, lifetime from $29.
Vitals
Activity Monitor alternative for Mac that groups processes by app, tracks coding agent sessions and AI usage limits, finds forgotten dev servers and exposes an MCP server; pay once, no subscription.
CybeDefend
Security and business rules for AI coding agents: one command installs VibeDefend into Claude Code, Cursor, Copilot or Codex to scan each diff and block risky actions, with SAST, SCA, secrets and IaC scanning; free plan.
The MCP Census
Checks public evidence for an MCP server and returns PASS, REVIEW, BLOCK or UNKNOWN before you install or call it; lookup, preflight, whole .mcp.json config audits, a change feed, REST API and its own MCP endpoint, no account needed for stats.
Skillsync
Mac app and CLI that searches all your past coding-agent sessions before each prompt and hands your agent a short recollection with sources; works with Claude Code, Codex, Cursor, OpenCode, Amp and more, indexes locally, with shared team workspaces.
PatchWave
GitHub app whose agent analyzes, fixes and merges Dependabot PRs using static analysis plus AI, with a prioritized inbox and auto-merge policies you control; 25 free dependency PRs a month, paid plans from $10/month.
MELRA
Open-source local MCP runtime that checks AI-agent tool calls against policy, gates risky actions on approval, verifies results and parks uncertain outcomes instead of blindly retrying after timeouts. Alpha; AGPL-3.0, SDKs Apache-2.0.
agent-archive
CLI to archive Claude Code, Codex and Cursor sessions to your own S3 or R2 bucket and hand a session's context off between agents or machines, plus browse history by topic, project, branch or PR. MIT-licensed, written in Go.
catnap
Claude Code plugin that warns Claude at 50% context so it finishes its current piece, writes itself a handoff note and compacts at a clean break instead of mid-task. MIT-licensed.
Omnara
Open-source managed agent platform, pitched as a self-hostable alternative to Claude Managed Agents and OpenAI's Agents API: run AI agents on any model and any machine through an API and CLI, with tools, secrets, permissions and streaming built in. Self-host free under Apache-2.0 or use Omnara Cloud.
SIwithin
Runs Claude Code or OpenAI Codex CLI unattended on a brief, fenced to the folders you give it and under a per-run dollar budget enforced by a loopback API proxy, leaving one ledger line per run. Single Python file, standard library only, macOS only, Apache-2.0.
Armite
Local Azure emulator with a real identity plane and RBAC: Entra tokens, managed identity, ARM, Key Vault and Blob Storage behind one RBAC engine that returns 403 where Azure would, driven unmodified by az, the Azure SDKs and Terraform. Apache-2.0.
Sumi.css
Minimalist, classless-first CSS framework: one file (6.7 KB min+gzip), no build step or JavaScript, automatic light/dark mode, theming via CSS custom properties and zero-specificity element styles that are easy to override. MIT-licensed.
agent-usage-guard
Claude Code mod that reads your plan's 5-hour and weekly usage and holds subagent fan-out, heavy-context prompts and retry loops before they spend your window, asking in Claude Code's own dialog. Local only, no network; MIT licensed.
Hookman
One permanent webhook URL that routes Stripe, Paddle and other provider events to whichever branch deployment is active, switched from a dashboard, CLI or GitHub Action. Runs on Cloudflare's edge; free tier covers 1,000 webhooks a month.
Slate Terminal
Native macOS terminal built on libghostty that your coding agents can control and extend: ask an agent to build custom panels, commands and layouts in HTML, CSS and JavaScript, and see which agent is working in each pane. Requires macOS 27 on Apple silicon.
Corgi
Open-source (MIT) terminal and agent multiplexer for macOS, Linux and WSL: shows what each Claude Code, Codex, Pi or OpenCode agent is doing and which needs you, keeps sessions running in the background like tmux, and tracks worktrees, changed files and per-workspace resource usage.
CATT
Consumer Agent Testing Tool: open-source (MIT) Python CLI that runs a scripted or model-driven surrogate assistant (Muse or Instinct emulation, unaffiliated) against your MCP server, MCP app, website or HTTP API, checks what the service actually did and records sealed evidence you can replay and diff in CI.
Lumabri
Open-source (Apache-2.0) tool written in C that runs a private inference service across your own computers on a LAN, splitting a model's layers between machines so models too big for one computer can run; powered by the Colibri engine.
ego lite
Free browser for AI agents on macOS, with Windows in beta: Claude Code, Codex or Cursor drive it through the open-source (MIT) ego-browser skill in their own isolated Spaces while you keep browsing, inheriting your Chrome logins and cookies; Linux is on the roadmap.
TypeUI Micro Interactions
Gallery of 113 interactive UI micro interactions (buttons, inputs, selection, modals, toasts, sliders, cards and more) you can preview and copy as implementation prompts for AI coding tools.
Enzyme
Open-source (Apache-2.0) Rust tool that builds a local index of your Markdown notes from their folders, tags, links and dates, so Claude Code, Codex or another terminal agent can pull up related notes and cite them.
proc
Open-source (MIT) Rust CLI for macOS, Linux and Windows that finds and manages processes by port, PID, name or path: see what is on :3000 and why, then stop or free it, with confirmations and --dry-run. Ships an agent skill so Claude Code, Codex, Cursor and Gemini CLI know when to use it.
pw-repl
Open-source (MIT) Playwright REPL that drives Chromium over CDP, in your own browser with its logins or a headless one, shared by you and your AI agents: agents send commands with pw-repl send, every command shows in the REPL, and watch records your manual steps for the agent to read.
skill-spotter
Open-source (MIT) Claude Code plugin that notices tasks and instructions you keep repeating and offers to save them as a /skill or a CLAUDE.md rule; data stays in a local SQLite file, no telemetry, and it never creates a skill without you.
QuickRust
Drop-in replacement for the cargo command from MachineFabric that speeds up Rust builds with nothing to configure: one central, size-limited build cache shared by every checkout, its own toolchains, and qrust analyze to show why crates rebuilt and where time and disk went.
Hob
Free, open-source (MIT) desktop app from Frigade for Mac, Windows and Linux that works on your files like the Claude and ChatGPT apps, but runs open-weight models (Gemma 4, gpt-oss) on your own computer via llama.cpp: picks a model for your hardware, writes sandboxed Python for ad hoc tools, reads images and PDFs, adds MCP connectors and works fully offline.
rostr
Open-source (MIT) CLI, hooks and skill that give every Claude Code or Codex session a unique name and emoji drafted from a roster you pick, so notifications and messages are traceable and agents can refer to each other; rostr stats shows who is working on what. Works with tmux and herdr; local by default, optional end-to-end encrypted cloud roster.
ADRs4AI
Method, repo template and VS Code extension for keeping the reasoning from human-AI work in Git: your agent turns your notes into Architectural Deliberation Records and asks open questions with options, a recommendation, rationale, confidence and a falsifier, which you accept or override; the extension's MCP server lets agents query the record.
Hatamoto
Open-source (Apache-2.0) batteries-included CMS that runs as one Go binary next to Postgres 16: visual page builder, SEO panel, page cache, forms, redirects, JSON-LD, sitemaps and llms.txt in core, server-side AI crawler analytics that flag spoofed bots, an MCP server with scoped keys, sandboxed WebAssembly plugins and a WordPress importer that audits every old URL. No telemetry, no paid edition.
Jog
Hosted GitHub Actions runners: change runs-on to jog and jobs run in a fresh VM on Jog's own hardware with Ubuntu 24.04 and GitHub's runner tools. Listed at $0.0012 a minute for 2 vCPU/8 GB, up to 32 vCPU/128 GB at $0.0176, with 5,000 free minutes a month while you have credits; works with personal GitHub accounts too.
Weir
Open-source (Apache-2.0) API key leak detection for AI agents: plants fake keys where real ones live, runs your agent's flows and reports whether they reached traces, logs, console output, the model or other services (leak, clean or incomplete), and in Claude Code shows which of your real keys entered or left a session, with an optional hook that redacts them. pip install weir-scan.
Ulak
Open-source (MIT or Apache-2.0) Rust CLI that runs a local Compose project's Docker workloads on a server you control over SSH, with no local Docker daemon: keeps your editor and checkout local, syncs the files Docker needs, runs Docker commands remotely and forwards published ports back to localhost. Installs nothing on the server; macOS and Linux.
Parley
Open-source (Apache-2.0) single Go binary that puts your Claude Code, Codex and OpenCode sessions on one local IRC channel with you, so they tell each other when something changes and ask when they need input; you read every line in any IRC client, only your nick carries authority, and a breaker stops runaway loops until you write.
Kavach
Open-source (Apache-2.0) flight recorder for journal-driven services: records every input, clock read and random read, and when a step panics or breaks an invariant writes a .kavach fixture that replays the crash byte for byte with no broker, database or network, then rejects narrow fixes that fail across perturbed variants. SDKs in 12 languages plus agent skills; pre-alpha.
Lasm
Open-source (MIT) compiler that turns ordinary Lean 4 programs into WebAssembly so one formally verifiable code base can ship as a Node server, AWS Lambda, Vercel, Netlify or Azure function, edge or WASI module, or a browser app, with Lean frameworks for HTTP, UI, typed RPC, DOM and KV plus JS/npm interop. Ships on npm with Lean bundled (npm create @lasm@latest); its RealWorld example carries 117 proved theorems. Early (0.3.x).
phpClaw
Open-source (MIT) AI agent framework for PHP 8.1+: a ReAct loop with tools (allowlisted shell, HTTP with SSRF checks, sandboxed file access, database queries, code search), memory and guard chains that screen messages before the model sees them. Works with Anthropic, OpenAI, Groq, Gemini, Mistral, DeepSeek or Ollama; eight adapters for Laravel, Symfony, Drupal, Magento, WordPress/WooCommerce, Joomla, PrestaShop and OpenCart, and an MCP server package that exposes your tools to Claude Code or Cursor.
CodeRabbit
AI code review for pull requests on GitHub, GitLab, Azure DevOps and Bitbucket, and before the PR in the CLI, your IDE and coding-agent loops: plain-language walkthroughs, one-click fixes, pre-merge checks, slop detection for low-quality AI PRs, and reviews that learn from your team's feedback. Free on public repos.
Greptile
AI code reviewer that indexes your repo as a graph of files, functions and dependencies, then runs a swarm of agents over each pull request to catch bugs beyond the diff, and learns your standards from teammates' review comments. Works with GitHub, GitLab and Bitbucket; Enterprise adds self-hosting, Gitea and Perforce.
Qodo
Multi-agent AI code review for pull requests on GitHub, GitLab, Bitbucket, Azure DevOps and Gerrit, with a rules system built from your PR history, cross-repo review that catches breaking changes downstream, and a CLI toolbox that gives Codex, Claude and Kiro a review step inside the agent loop.
Graphite
Code review platform for GitHub: stacked pull requests from a CLI, VS Code extension or MCP, a PR inbox, a merge queue, and AI reviews with custom rules, filters and one-click fixes. Free Hobby plan for personal repos.
Sourcery
AI code review for GitHub and GitLab, including self-managed GitLab: reviews every pull request with summaries, diagrams and one-click fixes, follows your custom review rules, runs security checks, and can auto-approve low-risk PRs. Free on public repos.
Bito
Bito's AI Code Review Agent: codebase-aware reviews on GitHub, GitLab and Bitbucket pull requests, in VS Code, JetBrains, Cursor and Windsurf, and from the CLI, with one-click fixes, custom guidelines and review analytics. Bito-hosted or self-hosted.
Firecrawl
Web data API for AI agents: scrape any URL into clean Markdown or JSON, crawl and map whole sites, search the web, run browser sessions and monitor pages, with an MCP server. Open source under AGPL-3.0 with a self-host guide, or use the cloud with 1,000 free credits a month.
Tavily
Search API built for AI agents: search, extract, map, crawl and research endpoints that return clean, LLM-ready results, with SDKs, an MCP server and a CLI. 1,000 free credits a month, no card.
Exa
Web search API for AI: real-time search with page contents, deep multi-step search, answers with citations, monitors and an async research agent, plus a hosted MCP server. Pay as you go, with $10 in free credits every month.
Apify
Web scraping and automation platform: thousands of ready-made scrapers (Actors) in the Apify Store plus your own, with proxies, storage and scheduling, and an MCP server so agents can run any Actor. Apify also maintains the open-source Crawlee library. Free plan with $5 of usage a month.
spinf-serve
Open-source (Apache-2.0) inference server for bulk classification and decisions on one GPU: it reads each piece of content once and scores many questions about it in the same pass, returning a probability over the options you give instead of generated text. OpenAI-compatible API, same as the hosted spinf service; needs Linux and a 24 GB NVIDIA GPU.
kube32
Open-source (Apache-2.0) Kubernetes-compatible control plane and WebAssembly pod runtime for ESP32-S3 microcontrollers: manage a cluster of boards with kubectl or k9s over Wi-Fi or USB serial, with pods written in Rust, C or AssemblyScript and a host simulator to try it without hardware.
capo-s3
Typed, async-first Python SDK for Amazon S3 and S3-compatible storage (RustFS, MinIO, Cloudflare R2), with asyncio and trio support, a matching sync client, multipart transfers and presigned URLs. Part of capo, an MIT-licensed community alternative to boto3 built on its own HTTP stack.
Mem0
Memory layer for AI agents and apps: send it conversations and it extracts, updates and retrieves the memories that matter for each user, through Python and Node SDKs, an API, MCP and a CLI. Open source (Apache-2.0) as a library or self-hosted server, or the hosted platform with a free plan.
Zep
Context and memory layer for AI agents: turns conversations, documents and business data into a temporal Context Graph per user, account or domain, with dated facts traced to their source and retrieval under 200 ms. Zep Cloud, your own encryption keys or your VPC; its graph engine Graphiti is open source (Apache-2.0). Free 10,000 credits a month.
Supermemory
Memory and context engine for AI agents: learns from each user's chats, files and URLs, stores it in a vector-graph database and injects what matters into your agent's context, through an API, MCP and plugins for Claude Code, Cursor and more. Open source (MIT); hosted free plan with $5 of credits a month, Pro $19/month.
Browserbase
Cloud browsers for AI agents: fleets of headless browsers that log in, fill forms and act where there is no API, plus Search and Fetch APIs, CAPTCHA solving, live view and session replay. Drive them with Playwright, Puppeteer, Selenium or Browserbase's open-source Stagehand SDK (MIT). Free plan; Developer $20/month.
Browser Use
Open-source (MIT) browser agent: run it from Python with any LLM, give Claude Code or Codex a browser through its CLI skill, or use Browser Use Cloud for stealth browsers with CAPTCHA solving and residential proxies and a hosted agent API. Pay as you go, no subscription: cloud browsers $0.02 an hour, $1 free credit.
Steel
Open-source (Apache-2.0) browser API for AI agents: Steel Browser runs Chrome sessions you drive with Puppeteer, Playwright or Selenium, keeps cookies and storage across requests, rotates proxies and turns pages into Markdown, screenshots or PDFs. Self-host it with Docker or use Steel Cloud, free to start with $30 of credits.
Playwright MCP
Microsoft's MCP server that lets Claude Code, Codex, Cursor, VS Code and other MCP clients drive a browser through Playwright's accessibility snapshots instead of screenshots: Chrome, Firefox, WebKit or Edge, headed or headless, or attached to your logged-in browser through an extension. Open source (Apache-2.0) and free.
LangGraph
LangChain's low-level orchestration framework for long-running, stateful agents: durable execution that resumes where a run failed, human-in-the-loop checks on agent state, and short- and long-term memory. Python 3.10+ or LangGraph.js, with or without LangChain. Open source (MIT); deploy it yourself or on LangSmith Deployment.
CrewAI
Open-source (MIT) Python framework for multi-agent automation: Crews of role-based agents that work one task together, and Flows for event-driven control that mix single LLM calls with Crews. Standalone, not built on LangChain. CrewAI AMP adds managed deployment, tracing and governance; free plan with 50 workflow executions a month.
OpenAI Agents SDK
OpenAI's open-source (MIT) SDK for multi-agent workflows in Python or TypeScript: agents with tools, handoffs, guardrails, sessions and human-in-the-loop, built-in tracing, sandbox agents for long tasks, and voice agents. Works with the OpenAI APIs and 100+ other LLMs. Free; you pay for model usage.
Mastra
TypeScript framework for AI agents and apps: agents with tools, graph-based workflows that suspend for human approval and resume, model routing across 40+ providers, built-in evals and observability, and Mastra Studio to test agents locally. Apache-2.0 outside its ee/ enterprise directories; Mastra Platform free to start, Teams $250/month.
Pydantic AI
Typed Python agent framework from the Pydantic team: any model a string swap away with fallback, validated structured outputs, MCP, evals and durable execution on Temporal, DBOS, Prefect and more. One agent runs as a CLI, a web chat or a realtime voice agent. Open source (MIT) and free; Logfire tracing is optional.
Google ADK
Google's open-source (Apache-2.0) Agent Development Kit: code-first agents and graph workflows, multi-agent hierarchies, MCP and OpenAPI tools, tool confirmation, a local dev UI and built-in evals. Optimized for Gemini but model-agnostic. Python, Java, Go, TypeScript and Kotlin; deploy with Docker, Cloud Run or Vertex AI Agent Engine.
Ghostman
Open-source (MIT) keyboard-driven terminal API client and webhook inspector built on Textual: compose GET, POST, PUT, PATCH and DELETE requests, see DNS and TTFB latency and formatted JSON, and catch incoming webhooks on a built-in local listener (port 9090) without ngrok. Local-first, no account; collections are plain JSON/YAML files you can commit to Git. Install with pip install ghostman (Python 3.10+).
Maclet
Open-source (Apache-2.0) container runtime for Apple Silicon Macs, written from scratch in Rust and Swift: instead of wrapping Docker Engine or containerd it boots a Linux VM through Apple's Virtualization Framework and implements namespaces, cgroups v2 limits, OverlayFS, bridge networking and OCI image pulls itself, with run, exec, logs, ps and stats commands. Early-stage project.
Charter
Open-source (Apache-2.0) Python library for AI agents that act in your customers' accounts through APIs: each tool is a Pydantic model that mirrors the API field by field with your rules on top, every call is validated before it is sent, the customer a call is for comes from your app rather than the model, and the same declarations produce a reviewable map of everything each tool can reach. Runs in your process with no proxy or telemetry; works with any HTTP API. pip install charter-ai.
Tinycap
Open-source (MIT) native macOS menu-bar screen recorder and screenshot tool for PR-sized clips: record a window, screen or dragged area, trim, and export MP4 (H.264 or HEVC), WebM or GIF with a 'fit under N MB' option, or mark up screenshots with arrows, text, pixelation and crop. Everything stays local. Requires macOS 15; install from the DMG or Homebrew.
CodeConnect
iPhone app plus a small daemon that puts Claude Code and Codex sessions from your Macs and Linux machines on one screen: see every agent waiting on you ranked by risk, approve or deny the exact command (risky ones need a hold and Face ID), answer questions, review diffs and take over the live terminal. Connects over your own Tailscale tailnet with no CodeConnect server or account. The app is free; iOS 17+, macOS 13+ or Linux x86_64/arm64.
ElevenLabs Agents
ElevenLabs' hosted platform for voice agents: its own speech-to-text, your choice of LLM or a custom model, 10,000+ voices, a turn-taking model, a visual workflow builder, knowledge bases, tests and A/B experiments. Deploys to phone over Twilio or SIP, a web widget, or iOS, Android, React Native and Unity SDKs. Free plan with 15 call minutes; $0.08 a minute after that, LLM billed on top.
Vapi
Developer platform for voice agents that make and take phone calls: pick the transcriber, model and voice from dozens of providers, build single assistants or Squads that hand callers between specialists, test with simulations and manage config as code from a CLI. $0.05 a minute for Vapi hosting, with providers passed through at cost and $5 of free credit.
Retell AI
Platform for AI phone agents: a drag-and-drop call-flow builder, simulation and regression tests, outbound campaigns, live call monitoring with takeover, IVR navigation and CRM write-back, plus chat and SMS agents. Pay as you go at $0.07 to $0.31 a minute depending on model and voice, 20 concurrent calls included and $10 of free credit.
Bland AI
Voice agent platform for phone calls with one per-minute rate that covers the LLM, speech-to-text and text-to-speech: Conversational Pathways for call flows, knowledge bases, version lock, your own Twilio or SIP trunk, and on-prem or VPC on Enterprise. Start plan $0.14 a minute with no platform fee and no card.
Synthflow
Enterprise platform for AI voice agents on customer calls, sold as one annual contract that covers telephony (Synthflow's own or SIP), CRM and calendar integrations, security review and launch by forward-deployed engineers. SOC 2 Type 2, ISO 27001, HIPAA and GDPR. No free plan or per-minute rate.
LiveKit Agents
Open-source (Apache-2.0) framework for realtime voice and video agents in Python or Node.js: mix any STT, LLM, TTS or realtime model, semantic turn detection, MCP tools, a built-in test framework and phone calls through LiveKit's SIP stack. Self-host the whole stack with the open-source LiveKit server, or deploy to LiveKit Cloud, free for 1,000 agent minutes a month.
Pipecat
Open-source (BSD-2-Clause) Python framework for realtime voice and multimodal agents, maintained by Daily: composable pipelines over WebRTC or WebSockets, 80+ LLM, STT and TTS services, Pipecat Flows for structured conversations, multi-agent handoffs and client SDKs from JavaScript to ESP32. Run it anywhere, or on Pipecat Cloud from $0.01 an active minute.
Together AI
Serverless API for open models such as gpt-oss, GLM and Kimi, OpenAI-compatible and billed per token, with batch, provisioned throughput, dedicated endpoints, fine-tuning and GPU clusters when you outgrow shared capacity.
Fireworks AI
Inference platform for open models: per-token serverless with Standard, Priority and Fast modes, batch at half the serverless price, on-demand GPUs billed per second, and LoRA fine-tuning with tuned models served at the base price. OpenAI-compatible.
Cerebras Inference
Open models such as GPT OSS 120B and Qwen 3.8 27B served on Cerebras hardware through an OpenAI-compatible API: a pay-as-you-go Developer tier for building and testing, Enterprise plans for production and fine-tuning.
Baseten
Per-token Model APIs for open models such as GLM, DeepSeek and Kimi, with OpenAI and Anthropic-compatible endpoints, plus dedicated deployments of your own or fine-tuned models packaged with the open-source Truss, billed by the minute.
DeepInfra
OpenAI-compatible API for open models such as Llama, DeepSeek and Qwen, billed per token with Standard, Priority and Flex tiers, plus dedicated GPUs for your own LLMs billed by the minute.
Modal Endpoints
Modal's inference product: OpenAI-compatible Shared Endpoints for open models billed per token, or Dedicated Endpoints that run open models or your own weights, billed per second of compute.
The 2 biggest AI stories, every morning.
One short email: the two stories that matter, and why.
You're in.
The Daily AI Brief will land in your inbox soon. If you don't see it, check Promotions or Spam and move it to your main inbox.