Here is how the pieces fit together as of 2026-04-17.
These are the tools developers use directly:
| Tool | Type | Model Access | Key Strength |
|---|---|---|---|
| Claude Code | CLI agent | Claude (Anthropic) | Deep project understanding, background agents |
| Cursor | IDE (VS Code fork) | Claude, GPT, custom | Inline editing, agent mode, .cursor/rules/ |
| GitHub Copilot | IDE extension | GPT, Claude | Ubiquitous, workspace agents |
| Windsurf | IDE (VS Code fork) | Multiple | Cascade agent, flow-based editing |
| Codex CLI | CLI agent | GPT (OpenAI) | Code generation from terminal |
| Continue | IDE extension | Any model | Open-source, model-agnostic |
| Aider | CLI agent | Any model | Git-native, multi-file editing |
For building custom AI agent applications:
| Framework | Language | Key Pattern |
|---|---|---|
| LangChain / LangGraph | Python, JS | Graph-based agent orchestration |
| CrewAI | Python | Multi-agent role-based collaboration |
| AutoGen | Python | Multi-agent conversation |
| Mastra | TypeScript | TypeScript-native agents |
| Semantic Kernel | C#, Python | Microsoft's enterprise agent framework |
| Vercel AI SDK | TypeScript | Streaming, tool use, React integration |
| Protocol | Purpose | Status |
|---|---|---|
| MCP | Connect agents to tools/data | Active — hundreds of servers in the registry |
| A2A | Agent-to-agent communication | Early stage |
| OpenAPI / Tool Use | Define callable functions | Mature |
| Model | Provider | Best For |
|---|---|---|
| Claude Opus 4.7 / Sonnet 4.6 | Anthropic | Agentic coding, long context (1M tokens) |
| OpenAI current flagship (see platform.openai.com/docs/models) | OpenAI | General coding, tool use |
| Gemini 3.1 Pro / Gemini 3 Flash | Multimodal, long context | |
| Codestral | Mistral | Open-source code generation |
| Llama (code-capable variants) | Meta | Local / private deployment |
Model lineups move fast. Re-verify against each vendor's pricing page before pinning versions in production.
Search for a command to run...