22 tool cards: use cases, strengths and weaknesses, pricing, alternatives.
OpenAI's flagship assistant. A universal tool for text, code, document analysis, voice input, and working with images. The most recognizable brand in the category, best supported by an ecosystem of plugins and custom GPTs.
Best for: Universal tasks: text, tech support, code, everyday questions, brainstorming. Good as a default 'second brain'.
Strengths
- A broad ecosystem and plugins
- Good writing quality
- Voice mode and image work
- Custom GPTs and projects with memory
Weaknesses
- Shorter context than Gemini/Claude
- Sometimes confidently wrong on facts
- A paid subscription for the best models
Pricing: Free tier with limits; ChatGPT Plus and Pro are paid·Difficulty: Beginner·Alternatives: claude, gemini, perplexity
Anthropic's assistant. Its strengths are a long context, careful work with large documents, reliable writing, careful code. Fewer 'hallucinations' in its cautious-phrasing mode.
Best for: Long documents, legal/financial text, careful fact-based answers, analytics, coding inside Claude Code.
Strengths
- A very long context
- Good writing and analysis quality
- Fewer confident errors
- Claude Code for development in the terminal
Weaknesses
- Fewer plugins and integrations
- A less developed voice mode
- Sometimes overly cautious in phrasing
Pricing: Free tier; Claude Pro and Max are paid·Difficulty: Beginner·Alternatives: chatgpt, gemini
Anthropic's terminal AI engineer. It works in a repository, writes code, runs tests, opens PRs, uses MCP servers and slash commands.
Best for: Development in existing repositories, refactoring, debugging, automating developer routine.
Strengths
- Direct access to the file system and shell
- MCP support
- Good code quality
- Headless mode for CI
Weaknesses
- Requires a Pro/Max subscription
- A learning curve for non-developers
Pricing: Included in Claude Pro / Max or the API·Difficulty: Intermediate·Alternatives: codex, cursor, windsurf
OpenAI's AI engineer: a CLI and cloud agents that work in a repository. Opens PRs, runs tasks in a sandbox, integrates with GitHub.
Best for: Long tasks in a sandbox, batch PRs, code review.
Strengths
- Cloud tasks without tying up your local machine
- Good GitHub integration
- Parallel tasks
Weaknesses
- Dependence on the ChatGPT infrastructure
- Less flexible local-environment settings
Pricing: Part of ChatGPT Plus / Pro / Team·Difficulty: Intermediate·Alternatives: claude-code, cursor
A framework for building multi-agent systems. A convenient abstraction of a 'team of roles' that communicate with each other.
Best for: Experiments with multi-agent scenarios, demos, prototypes.
Strengths
- A simple 'roles + tasks' model
- Good documentation
Weaknesses
- Less control than LangGraph
- Sometimes overcomplicates things in production
Pricing: Open source; paid CrewAI+·Difficulty: Intermediate·Alternatives: langgraph, flowise
A VS Code-based AI IDE with deep model integration. Suited for those who want a familiar editor + a strong AI assistant right inside it.
Best for: Pair programming with the model, refactoring, LLM-level autocomplete.
Strengths
- The familiar VS Code interface
- Support for several models
- Good work with diffs
Weaknesses
- A paid subscription for the best features
- Sometimes overdoes the autocomplete
Pricing: Free with limits; Pro is paid·Difficulty: Intermediate·Alternatives: windsurf, claude-code
A Chinese open-weight model lineup. Known for strong reasoning at a low price. Available both via an API and for local runs.
Best for: Reasoning tasks, code, math. A good choice when you need a budget API.
Strengths
- Strong reasoning
- Low API price
- Open weights — can be run locally
Weaknesses
- Regional restrictions and compliance
- A less developed ecosystem
Pricing: Cheap API; open weights are free·Difficulty: Intermediate·Alternatives: qwen, mistral, claude
An open-source visual builder of LLM apps and agents. An analog of Langflow.
Best for: Prototypes of agents and RAG pipelines without code.
Strengths
- A visual no-code builder
- Self-host
- Good for demos
Weaknesses
- Usually replaced with code in production
Pricing: Open source; paid Cloud·Difficulty: Intermediate·Alternatives: langgraph, crewai
Google's model lineup. Strong in multimodality (images, audio, video), in integration with Google Workspace, and in tasks with a very long context (up to a million tokens).
Best for: Work inside the Google ecosystem, analysis of long documents and video, multimodal tasks.
Strengths
- A huge context (up to 1M tokens)
- Integration with Gmail, Docs, Drive
- Good multimodality
- Access to Google Search
Weaknesses
- Writing quality sometimes lags behind Claude/GPT
- Behavior changes between versions
Pricing: Free tier; Gemini Advanced is paid·Difficulty: Beginner·Alternatives: chatgpt, claude
An xAI model with a focus on realtime data from X (Twitter). Less regulated than the mainstream, with a 'less polished' answer style.
Best for: Fresh news, realtime context from X, topics where other models bail out.
Strengths
- Realtime access to the X feed
- Less conservative phrasing
- Good speed
Weaknesses
- Fewer mature integrations
- Access via X Premium
- Lower code quality than the flagships
Pricing: Access via an X Premium subscription / the xAI API·Difficulty: Intermediate·Alternatives: chatgpt, perplexity
A desktop app for running LLMs locally with a UI. A convenient model catalog, chat, and server.
Best for: Those who need a GUI around local models.
Strengths
- A convenient GUI
- A model catalog inside
- An OpenAI-compatible local server
Weaknesses
- Desktop only
- A less server-oriented workflow than Ollama
Pricing: Free·Difficulty: Beginner·Alternatives: ollama
A LangChain framework for building agents as state graphs. Suited when you need explicit control over transitions and retry logic.
Best for: Production agents with complex branching and long loops.
Strengths
- Explicit control over state
- Maps well onto complex processes
- LangSmith for tracing
Weaknesses
- A learning curve
- Overkill for simple tasks
Pricing: Open source; LangSmith is paid·Difficulty: Advanced·Alternatives: crewai, flowise
A visual automation platform (formerly Integromat). The most convenient for those who want to see the pipeline 'with their own eyes' without code.
Best for: Non-technical teams, marketing pipelines, medium-complexity scenarios.
Strengths
- A very clear visual editor
- Thousands of ready-made modules
- A good price per operation
Weaknesses
- Less flexible than n8n
- Cloud-only
Pricing: Free tier with limits; paid plans·Difficulty: Beginner·Alternatives: n8n, zapier
A French company and model lineup. Open-weight Mistral / Mixtral models and paid Le Chat and API. Strong in European languages and privacy.
Best for: Local runs, European languages, privacy-sensitive cases.
Strengths
- Open weights
- Good European languages
- EU compliance
Weaknesses
- Lower quality than the flagships
- A less developed plugin ecosystem
Pricing: API and Le Chat; open weights are free·Difficulty: Intermediate·Alternatives: deepseek, qwen
A Google tool for working with your own corpus of documents. You upload PDFs, text, video — and get a chat, summaries, audio overviews of your material.
Best for: Research, studying long documents, exam prep, notes.
Strengths
- Cites its own sources
- Podcast-format audio overviews
- Free for most cases
Weaknesses
- Limits on source size
- Tied to a Google account
Pricing: Free; Plus as part of Google AI Pro·Difficulty: Beginner·Alternatives: claude, perplexity
The simplest way to run an LLM locally. One binary, the command `ollama run llama3` — and the model is running.
Best for: Running models locally, experiments, privacy-sensitive scenarios.
Strengths
- Very simple to run
- An OpenAI-compatible API
- A large model catalog
Weaknesses
- Fewer fine settings than llama.cpp
- Depends on hardware
Pricing: Free (open source)·Difficulty: Intermediate·Alternatives: lm-studio, openrouter
A single API router to dozens of models from different providers. One key — you pick the model in the request.
Best for: Experiments with different models without separate subscriptions, production with auto-fallback.
Strengths
- One key — many models
- Transparent prices
- Fallback between models
Weaknesses
- Sometimes more expensive than direct APIs
- Not all exclusive features are available
Pricing: Pay-as-you-go·Difficulty: Intermediate·Alternatives: ollama
An AI search engine. It does what Google does poorly: gives a direct answer with links to sources. Well suited for research, fact-checking, and comparisons.
Best for: Search with source citation, research, fact-checking, comparing products.
Strengths
- Links to sources
- Good at finding fresh information
- A convenient Pro Search mode
- Spaces for topic-based contexts
Weaknesses
- Less suited for generating long text
- Quality depends on the quality of the sources
Pricing: Free tier; Pro and Max are paid·Difficulty: Beginner·Alternatives: chatgpt, gemini
Alibaba's model family. Strong in multilingualism, code models, and vision. Many versions in open weights.
Best for: Multilingual tasks, local runs, vision tasks.
Strengths
- Strong multilingualism
- Code models in open weights
- Active development
Weaknesses
- Writing quality lags behind GPT/Claude in some languages
- Regional nuances
Pricing: API; open weights are free·Difficulty: Intermediate·Alternatives: deepseek, mistral
An AI IDE focused on agent mode: the model walks the project itself, makes changes, runs commands. An alternative to Cursor.
Best for: Agentic tasks in the IDE, long changes across multiple files.
Strengths
- A strong agent mode (Cascade)
- Transparent agent steps
Weaknesses
- Fewer users and plugins than Cursor
Pricing: Free with limits; paid tiers·Difficulty: Intermediate·Alternatives: cursor, claude-code
The oldest no-code automator. A huge number of integrations out of the box, built-in AI actions.
Best for: Simple 'when X — do Y' scenarios, fast SaaS integration.
Strengths
- A huge number of integrations
- AI Actions without code
- A very friendly UX
Weaknesses
- Expensive for a large volume of tasks
- Less flexibility in complex scenarios
Pricing: Free tier; paid plans by task count·Difficulty: Beginner·Alternatives: make, n8n
A self-hostable automation platform with a visual editor. The most flexible of the 'no-code' tools — it lets you insert code, use LLM nodes, and connect anything via HTTP.
Best for: Complex automations where flexibility and control over the infrastructure matter.
Strengths
- Self-host (privacy)
- Flexible code nodes
- Ready-made AI nodes for various providers
- A large library of integrations
Weaknesses
- Requires an understanding of requests/JSON
- The cloud version is more expensive than Make
Pricing: Self-host is free; Cloud is paid·Difficulty: Intermediate·Alternatives: make, zapier