02The tool map / Every tool

601 tools across applied AI.

Eleven areas, from the data a model learns from to the apps built on it and the landscape around them.

I want to build

Emerging now.

Launched in the last six months, newest first. Tap a tag to see why.
  • PiCoding assistants
    122kPopularQ4 2026

    Minimal, extensible terminal coding agent and agent toolkit from Earendil

    MITearlyself-host
  • Reflection BeamOpen-weight models
    Q4 2026

    501B MoE open weight reasoning model from Reflection AI, 23B active, 1M context

    Apache-2.0earlycloud or self-host
  • Q4 2026

    Open weight speech LLM for transcription in 40 languages, diarization and translation

    OpenMDW-1.1earlyself-host
  • Q4 2026

    Mistral transcription and TTS models plus a voice agent SDK built on LiveKit

    Proprietary (preview)earlycloud
  • Q4 2026

    Connecticut AI law: provenance data, frontier whistleblower protection, AEDT duties

    Regulation
  • OpenAI Agents APIAgent platforms
    Q3 2026

    Managed API that runs the Codex harness with hosted sandboxes for long running agents

    Proprietaryearlycloud
  • 282kPopularQ3 2026

    Plugin based agent harness for coding agents around DeepSeek models, with a web UI

    MITearlyself-host
  • Q3 2026

    Requires large GenAI providers to offer AI detection tools and provenance disclosures

    Regulation
  • 66Q3 2026

    MCP extension for long running, pollable tool calls, split from core in 2026-07-28

    MITearlycloud or self-host
  • Vercel eveFrameworks
    5.9kQ2 2026

    TypeScript agent framework where each agent is a directory compiled into a service

    Apache-2.0earlycloud or self-host
  • OpenAI SitesApp builders
    Q2 2026

    Codex plugin that builds and deploys OpenAI hosted websites, web apps and games

    Proprietaryearlycloud
  • Unlimited OCRDocument AI
    31.2kPopularQ2 2026

    Baidu OCR model for one shot parsing of very long documents, a DeepSeek-OCR successor

    MITearlyself-host

Most starred on GitHub.

Open-source tools on the map, by GitHub stars on 2026-11-16.
  1. 01OpenClawopenclaw/openclaw395k
  2. 02DeepSeek Harnessdeepseek-ai/deepseek-harness282k
  3. 03Hermes Agentnousresearch/hermes-agent264k
  4. 04OpenCodeanomalyco/opencode227k
  5. 05n8nn8n-io/n8n211k
  6. 06Firecrawlfirecrawl/firecrawl202k
  7. 07Ollamaollama/ollama187k
  8. 08Hugging Face Transformershuggingface/transformers169k
  9. 09Difylanggenius/dify161k
  10. 10Open WebUIopen-webui/open-webui159k

601 tools

Data.

Labeling, pipelines, catalogs and lineage, synthetic data and quality: what every model is built on.

Labeling 6
  • CVATLabeling

    Annotation tool for images, video and 3D point clouds in computer vision projects

    MITwidely usedcloud or self-host
  • EncordLabeling

    Curate, annotate and evaluate multimodal data incl. video, DICOM and LiDAR

    Proprietaryproduction-readycloud
  • 28.7kPopular

    Open source labeling tool for text, images, audio, video and time series

    Apache-2.0widely usedcloud or self-host
  • LabelboxLabeling

    Labeling platform plus managed labeling services for RLHF and evaluation data

    Proprietarywidely usedcloud
  • Scale AILabeling

    Data labeling and expert data services for training and evaluating models

    Proprietarywidely usedcloud
  • Snorkel AILabeling

    Programmatic labeling, expert datasets and evaluation for specialized AI

    Proprietaryproduction-readycloud or self-host

Models.

Model providers, open-weight families and the hubs that host them.

Open-weight models 11
  • Reflection BeamOpen-weight models
    Q4 2026

    501B MoE open weight reasoning model from Reflection AI, 23B active, 1M context

    Apache-2.0earlycloud or self-host
  • DeepSeekOpen-weight models
    105kPopular

    Open weight LLMs (V3 and R1 lines) plus a low-cost API

    MITwidely usedcloud or self-host
  • GemmaOpen-weight models

    Google's open weight models sized for single GPUs and on-device use

    Gemma Terms of Useproduction-readyself-host
  • GLM (Z.ai)Open-weight models

    Z.ai's open weight GLM models for reasoning, coding and agents

    MITproduction-readycloud or self-host
  • gpt-ossOpen-weight models
    20.5kPopular

    OpenAI's open weight reasoning models in 120b and 20b sizes

    Apache-2.0production-readyself-host
  • Kimi K2Open-weight models

    Moonshot AI's open weight 1T-parameter MoE models for agentic and coding work

    Modified MITproduction-readycloud or self-host

Training.

Fine-tuning, distributed training, experiment tracking and AutoML.

Fine tuning 7
  • AxolotlFine tuning

    Config-driven post-training and fine-tuning framework for LLMs

    Apache-2.0production-readyself-host
  • LLaMA-FactoryFine tuning
    76.1kPopular

    Fine-tuning for 100+ LLMs and VLMs via CLI or web UI

    Apache-2.0production-readyself-host
  • PEFTFine tuning
    21.9kPopular

    Parameter-efficient fine-tuning methods such as LoRA and QLoRA

    Apache-2.0widely usedself-host
  • TinkerFine tuning

    Managed API from Thinking Machines for LoRA fine-tuning and RL of open models

    Proprietaryearlycloud
  • TRLFine tuning

    Hugging Face library for SFT, DPO, GRPO and other post-training methods

    Apache-2.0widely usedself-host
  • UnslothFine tuning
    79.6kPopular

    Faster, lower-memory fine-tuning and RL for open LLMs

    Apache-2.0production-readyself-host

Inference.

Serving engines, GPU clouds, edge runtimes, gateways and caching.

Deployment 18
  • AIBrixDeployment

    vLLM project control plane for scalable LLM serving on Kubernetes

    Apache-2.0earlyself-host
  • Serverless A100 and T4 GPUs for containers with per-second billing

    Proprietaryproduction-readycloud
  • BeamDeployment

    Serverless GPU runtime whose open-source engine beta9 can be self-hosted

    AGPL-3.0production-readycloud or self-host
  • CerebriumDeployment

    Serverless GPU platform for real-time AI apps with per-second billing

    Proprietaryproduction-readycloud
  • CogDeployment

    Replicate open-source tool that packages models into standard containers

    Apache-2.0production-readyself-host
  • dstackDeployment

    Orchestrates AI dev, training and inference across clouds, Kubernetes and on-prem

    MPL-2.0production-readycloud or self-host

Retrieval.

Vector databases, search, embeddings, rerankers and document parsing.

Vector databases 14
  • ChromaVector databases
    29.7kPopular

    Embedded single-node vector store with a Rust core

    Apache-2.0production-readycloud or self-host
  • FAISSVector databases
    41.7kPopular

    ANN index library rather than a database; baseline for offline search

    MITwidely usedself-host
  • LanceDBVector databases

    Serverless vector database on the Lance columnar format, reads from object storage

    Apache-2.0production-readycloud or self-host
  • Milvus / ZillizVector databases
    46.7kPopular

    Distributed billion-scale vector database; an LF AI & Data project

    Apache-2.0widely usedcloud or self-host
  • Vector index inside MongoDB with native embedding and rerank APIs

    Commercial / SSPL corewidely usedcloud
  • pgvectorVector databases
    23.7kPopular

    Postgres extension: HNSW/IVFFlat vector indexes in your existing database

    PostgreSQLwidely usedself-host

Agents.

Frameworks, protocols, context management, memory, computer use and sandboxes.

Frameworks 50
  • 282kPopularQ3 2026

    Plugin based agent harness for coding agents around DeepSeek models, with a web UI

    MITearlyself-host
  • Vercel eveFrameworks
    5.9kQ2 2026

    TypeScript agent framework where each agent is a directory compiled into a service

    Apache-2.0earlycloud or self-host
  • AG2Frameworks

    Community fork of AutoGen 0.2 for conversational multi agent systems

    Apache-2.0earlyself-host
  • Agency SwarmFrameworks

    Framework for agent teams with defined roles, built on the OpenAI Agents SDK

    MITearlyself-host
  • AgentScopeFrameworks
    36.1kPopular

    Multi agent framework with tools, memory and visual debugging, from Alibaba

    Apache-2.0earlyself-host
  • AgnoFrameworks
    43.2kPopular

    Python framework and runtime for multi-agent systems with memory and tools

    Apache-2.0production-readycloud or self-host

Evaluation.

Evals, tracing, observability and drift monitoring.

Tracing 23
  • AgentOpsTracing

    Agent observability SDK with session replays across agent frameworks

    MITfragilecloud or self-host
  • OTel-native tracing and eval workflow that runs fully local

    Elastic-2.0widely usedcloud or self-host
  • Eval-first workflow with scoring, prompt playground and a VPC data plane

    Proprietarywidely usedcloud or self-host
  • 23.4kPopular

    Tracing, evaluation and guardrail hooks with no enterprise feature gating

    Apache-2.0production-readycloud or self-host
  • LLM spans inside existing APM, correlated with infrastructure telemetry

    Proprietarywidely usedcloud
  • HeliconeTracing

    Proxy-based logging and caching with cache-hit-rate dashboards

    Apache-2.0production-readycloud or self-host

Governance.

Guardrails, AI security, privacy and the rules that apply.

Guardrails 19
  • Managed content filters and grounding checks; ApplyGuardrail works with any model

    Proprietarywidely usedcloud
  • Arthur EngineGuardrails

    Open-source real-time evaluation and guardrails engine for GenAI apps

    MITearlycloud or self-host
  • Managed moderation and prompt-injection filters

    Proprietaryproduction-readycloud
  • Edge WAF detections that score prompt injection and flag unsafe LLM prompts

    Proprietaryproduction-readycloud
  • Google Cloud service screening prompts and responses for injection, data loss and bad URLs

    Proprietaryproduction-readycloud
  • OpenAI open-weight reasoning models that classify content against your own policy

    Apache-2.0earlyself-host

Modalities.

Speech, vision, document AI, and image and video generation.

Realtime and voice APIs 20
  • Agora Conversational AI EngineRealtime and voice APIs

    Voice agent engine on Agora's RTC network with pluggable LLM and TTS

    Proprietaryproduction-readycloud
  • Amazon Nova SonicRealtime and voice APIs

    Bedrock speech-to-speech models with a bidirectional streaming API

    Proprietaryproduction-readycloud
  • Azure Voice Live APIRealtime and voice APIs

    Unified Azure speech-to-speech API combining STT, generative models and TTS

    Proprietaryproduction-readycloud
  • Bland AIRealtime and voice APIs

    Enterprise phone agent platform with its own telephony, STT, LLM and TTS stack

    Proprietaryproduction-readycloud
  • DailyRealtime and voice APIs

    WebRTC infrastructure and SDKs used to transport realtime voice agents

    BSD-2-Clause (SDK)widely usedcloud
  • Deepgram Voice Agent APIRealtime and voice APIs

    Single voice-to-voice API combining Nova STT, Aura TTS and LLM orchestration

    Proprietaryproduction-readycloud

Builders.

AI coding assistants, app builders, notebooks and AI IDEs.

Agent UI kits 27
  • A2UIAgent UI kits

    Generative UI format: agents stream declarative layouts that apps render natively

    Apache-2.0earlyself-host
  • AnythingLLMAgent UI kits
    68.3kPopular

    Desktop and Docker app for private chat with documents, agents and MCP tools

    MITproduction-readycloud or self-host
  • assistant-uiAgent UI kits

    React primitives for AI chat: threads, streaming, tool call UIs, attachments

    MITproduction-readyself-host
  • ChainlitAgent UI kits

    Python framework for chat and agent UIs with step and tool call tracing

    Apache-2.0production-readyself-host
  • Chatbot UIAgent UI kits
    33.3kPopular

    Open source ChatGPT style web UI for multiple model providers

    MITfragileself-host
  • Deep ChatAgent UI kits

    Framework agnostic chat web component that connects to many AI APIs

    MITproduction-readyself-host

Landscape.

See what others build with, compare models and prices, track the field, and meet the foundations that steward open standards.

Model benchmarks 6
  • Artificial AnalysisModel benchmarks

    Independent benchmarks of model quality, speed, latency and price

    Proprietarywidely usedcloud
  • Public database of model benchmark results and trend data, downloadable

    CC-BY-4.0production-readycloud
  • LMArena (Arena)Model benchmarks

    Crowdsourced head-to-head model comparisons with Elo-style leaderboards

    Proprietarywidely usedcloud
  • OpenRouter RankingsModel benchmarks

    Model and app usage rankings by tokens processed on OpenRouter

    CC-BY-4.0production-readycloud
  • Scale SEAL LeaderboardsModel benchmarks

    Expert-run private evaluations of frontier models, incl. SWE-Bench Pro

    Proprietaryproduction-readycloud
  • SWE-benchModel benchmarks

    Benchmark and leaderboard of models fixing real GitHub issues

    MITwidely usedcloud
My daily stack

The stack I use, every day.

Thirty tools I use daily. Tap a bead to see what each one is for.

Applied AI
Data & retrieval
Product engineering
Applied AI

Claude Code.

Anthropic's coding agent in the terminal and editor.

Where I use it: Daily; Tokalator was built to budget its context

www.anthropic.com/claude-code