Independent AI model and skill directory

Choose the right AI stack with less guesswork.

VerdictLens helps teams browse AI models, supporting tools, and practical use cases with clear trade-offs, official links, and structured data they can reuse.

01
29 verified AI models
02
1528 live skills with setup notes and official links
03
28 editor-reviewed picks · 547 verified listings · 953 broader directory entries

Quick shortlist snapshot

Already have finalists? Compare 2–3 side by side.

Use compare after browsing models to pressure-test pricing, speed, context window, and workflow fit in one table.

GPT-5
OpenAI
Score
92
Claude Sonnet 4.6
Anthropic
Score
92
Gemini 2.5 Pro
Google
Score
91
Open compare workspace

Start here

Three clear ways to start.

Browse by model, browse by supporting tools, or start from the job you need done. The site is now organized around those three decisions first.

Featured models

Browse AI models

Browse leading models with clear pricing context, speed, strengths, and visible official links.

OpenAI logo

OpenAI

GPT-5

Overall score
92

OpenAI’s previous flagship reasoning model for coding, agentic tasks, and broad professional work.

reasoningcodingagentic
Context window
400K tokens
Speed
Balanced
Anthropic logo

Anthropic

Claude Sonnet 4.6

Overall score
92

Anthropic’s balanced Claude tier for broad production use, coding, and agent orchestration.

balancedcodingagents
Context window
200K tokens
Speed
Balanced
Google logo

Google

Gemini 2.5 Pro

Overall score
91

Google’s most advanced Gemini 2.5 model for complex reasoning and coding tasks.

multimodalreasoninglong-context
Context window
1M tokens
Speed
Balanced
Perplexity logo

Perplexity

Sonar Pro

Overall score
87

Perplexity’s advanced search model with deeper content understanding and enhanced search accuracy.

searchresearchgrounded
Context window
Search-context dependent
Speed
Balanced

Featured skills

Browse AI skills and tools

The tool layer often determines whether a model stays dependable in real work. These picks make that layer easier to inspect.

Coding & devtools · CLI coding agent

Codex CLI

Overall score
90

OpenAI’s terminal-first coding agent for editing code, running commands, and agentic development loops.

codingcliagent
Difficulty
Easy
Source
OpenAI docs

Memory & knowledge · Workspace knowledge access

Notion MCP

Overall score
86

Notion’s official MCP integration for exposing workspace search, pages, comments, and structured data sources to AI clients.

mcpnotionknowledge
Difficulty
Moderate
Source
Notion docs

Execution & sandboxes · Containerized MCP gateway

Docker MCP Gateway

Overall score
85

Docker’s gateway for running and brokering MCP servers in containerized, easier-to-govern environments.

dockermcpgateway
Difficulty
Moderate
Source
Docker docs

Coding & devtools · Agent skill installer and discovery CLI

Vercel Skills

Overall score
86

Vercel’s open skills CLI for discovering, installing, and managing reusable agent skills across multiple coding agents.

skillscliagentskills
Difficulty
Easy
Source
Vercel docs

Use cases

Use-case guides

Start with the job to be done, then narrow down the model and skill mix that fits your workflow.

How scoring works

Structured enough to trust. Flexible enough to keep improving.

Model score = capability 30, use-case fit 25, cost efficiency 15, speed 10, reliability 10, agent readiness 10.

Skill score = utility 25, compatibility 20, ease of setup 15, reliability 15, docs quality 10, adoption 10, safety & maintenance 5.

Scores combine benchmark signals, product experience, and editorial weighting. Use them as a practical guide, not an absolute truth claim.

Structured data for teams and agents

Built with clean public data from day one.

Each endpoint is easy to inspect, reuse, or index—useful for websites, internal tooling, search, and AI answer engines.