本页目录6 个章节
Editorial snapshot from September 2026. Prices, models and interface labels may change; this is a practical guide, not a live product feed.
Compare the environment, billing model and verified capabilities. Keep IDEs, AI platforms and coding agents in their own categories.
All tables use the same dated tool records as the directory, finder and pricing pages. A missing capability is an evidence gap, not a negative rating.
AI-native IDEs & native editors
Verified capabilities. “Not verified” is an evidence gap, not a claim that a feature is unavailable. Features can depend on plan, platform and selected agent.
| Decision | TraeCode 2026-09-21 | Cursor 2026-09-21 | Kiro 2026-09-21 | Google Antigravity 2026-09-21 | Zed 2026-09-21 | Devin Desktop 2026-09-21 | Qoder IDE 2026-09-21 | CodeBuddy IDE 2026-09-21 |
|---|---|---|---|---|---|---|---|---|
| Environment | AI-native IDE | AI-native IDE | AI-native IDE | AI-native IDE | Native editor | AI-native IDE | AI-native IDE | AI-native IDE |
| Platforms | macOS, Windows, Linux | macOS, Windows, Linux, Web | macOS, Windows, Linux, Web | macOS, Windows, Linux | macOS, Windows, Linux | macOS, Windows, Linux | macOS, Windows, Linux | macOS, Windows |
| Good starting point for | Developers who want code navigation and an agent in the same workspace. | Teams evaluating an integrated editor and hosted agent workflow. | Projects where requirements and implementation tasks benefit from a written plan. | Developers exploring parallel agent tasks and browser-assisted development. | Developers who want to choose their model provider or connect an existing coding agent. | Existing Windsurf users and developers coordinating several agents from one surface. | Developers who want to explore a repository and delegate changes from their editor. | Developers evaluating an integrated editor for chat, completion and multi-file tasks. |
| Pricing model | Subscription + metered model usagePaid from $20/mo | Subscription + model usage poolsPaid from $20/mo | Subscription + work creditsPaid from $20/mo | Free tier + Google plan allowancesFree access; see limits | Free editor / Pro + token usagePaid from $10/mo | Subscription + plan-dependent usagePaid from $20/mo | Check current subscription and usage termsSee official pricing | Subscription + creditsPaid from $10/mo |
| Free access | Available; limits apply | Available; limits apply | Available; limits apply | Available; limits apply | Available; limits apply | Available; limits apply | Not verified | Available; limits apply |
| Model access | The official model list is dynamically rendered. Check the in-app picker; this guide does not promise specific model versions. | Hosted models; availability and rates vary by model. | Hosted models with plan, regional and credit-multiplier restrictions. | Hosted Gemini and Claude models; the pricing page also lists gpt-oss, which is not treated as the GPT family here. | Hosted, provider API, external-agent or local access; these are distinct configurations. | Agent selection and model selection are different. Consult the provider for the chosen agent. | Check the current IDE model selector and plan; specific model families are not verified here. | Consult the IDE model selector and current credit rules; availability can differ by account and region. |
| Agent workflow | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified |
| CLI surface | Not verified | Verified | Verified | Verified | Not verified | Not verified | Not verified | Not verified |
| MCP | Verified | Verified | Verified | Not verified | Verified | Verified | Not verified | Not verified |
| Skills | Verified | Verified | Not verified | Not verified | Verified | Not verified | Not verified | Verified |
| Rules / instructions | Verified | Not verified | Verified | Not verified | Verified | Not verified | Not verified | Not verified |
| Hooks | Not verified | Verified | Verified | Not verified | Not verified | Not verified | Not verified | Not verified |
| Memory | Verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified |
| Bring your own key | Not verified | Not verified | Not verified | Not verified | Verified | Not verified | Not verified | Not verified |
| Local models | Not verified | Not verified | Not verified | Not verified | Verified | Not verified | Not verified | Not verified |
| Cloud agents | Not verified | Verified | Not verified | Not verified | Not verified | Verified | Not verified | Not verified |
| Parallel agents | Not verified | Not verified | Not verified | Verified | Verified | Verified | Not verified | Not verified |
| Check before switching | Usage is metered. Check the model picker and regional plan before moving an existing workflow. | A subscription is not unlimited frontier-model usage. Track included pools and on-demand spend. | Credits measure work rather than a fixed number of prompts. Model choice changes consumption. | Free access has rate limits. Paid benefits depend on your Google plan and account. | External agents have their own authentication and billing; an editor subscription does not cover every agent. | Check the migration FAQ for your existing plan; legacy entitlements can differ from new subscriptions. | Qoder also offers desktop, CLI and plugin products. Check the IDE download and its current account limits specifically. | Credit bonuses can be promotional. Compare the base allowance and model consumption before relying on the advertised total. |
Established IDEs with AI
Verified capabilities. “Not verified” is an evidence gap, not a claim that a feature is unavailable. Features can depend on plan, platform and selected agent.
| Decision | VS Code + AI 2026-09-21 | JetBrains IDEs + AI 2026-09-21 | Visual Studio + Copilot 2026-09-21 | Positron 2026-09-21 |
|---|---|---|---|---|
| Environment | IDE with AI integration | IDE with AI integration | IDE with AI integration | IDE with AI integration |
| Platforms | macOS, Windows, Linux, Web | macOS, Windows, Linux | Windows | macOS, Windows, Linux |
| Good starting point for | Developers keeping an existing extension, debugging and source-control setup. | Java, Kotlin and Python projects that depend on language-aware refactoring, debugging and project tooling. | Windows, C# and .NET development that relies on Visual Studio’s solution and debugging tools. | Analysts and developers combining interactive data exploration with agent-assisted coding. |
| Pricing model | Free editor + separate AI provider planFree access; see limits | IDE license + AI plan or external provider usageSee official pricing | Visual Studio edition + separate Copilot planSee official pricing | Free editor + separate model-provider usageFree access; see limits |
| Free access | Available; limits apply | Not verified | Not verified | Available; limits apply |
| Model access | Depends on the installed agent and account; do not treat the editor as a model subscription. | AI Assistant supports external providers and local models. Feature availability depends on the IDE, model and agent integration. | Copilot model access depends on the selected plan and Visual Studio integration. | Connect a supported provider with its own account or API key. Provider billing is separate from the editor. |
| Agent workflow | Verified | Verified | Verified | Verified |
| CLI surface | Not verified | Not verified | Not verified | Not verified |
| MCP | Verified | Verified | Not verified | Verified |
| Skills | Verified | Not verified | Not verified | Verified |
| Rules / instructions | Verified | Verified | Not verified | Not verified |
| Hooks | Verified | Not verified | Not verified | Not verified |
| Memory | Not verified | Not verified | Not verified | Verified |
| Bring your own key | Not verified | Verified | Not verified | Verified |
| Local models | Not verified | Verified | Not verified | Not verified |
| Cloud agents | Not verified | Not verified | Not verified | Not verified |
| Parallel agents | Verified | Not verified | Not verified | Not verified |
| Check before switching | Editor support does not imply identical capabilities across agents. Account access and billing belong to the selected provider. | AI features vary by IDE and edition. An IDE license and a model-provider subscription can be separate costs. | This is Visual Studio, a separate product from VS Code. Check IDE edition eligibility and Copilot billing independently. | The editor is free; connected AI providers have their own account requirements and charges. |
AI platforms and assistants
Verified capabilities. “Not verified” is an evidence gap, not a claim that a feature is unavailable. Features can depend on plan, platform and selected agent.
| Decision | GitHub Copilot 2026-09-21 | JetBrains Junie 2026-09-21 |
|---|---|---|
| Environment | AI platform / assistant | AI platform / assistant |
| Platforms | macOS, Windows, Linux, Web | macOS, Windows, Linux |
| Good starting point for | Teams whose code review and pull-request workflow already lives on GitHub. | Developers who rely on JetBrains project understanding and language tooling. |
| Pricing model | Subscription + GitHub AI creditsPaid from $10/mo | JetBrains AI plans; check regional pricingSee official pricing |
| Free access | Available; limits apply | Not verified |
| Model access | Hosted model selection varies by plan and environment. | Check current Junie settings and JetBrains AI plan documentation. |
| Agent workflow | Verified | Verified |
| CLI surface | Verified | Not verified |
| MCP | Verified | Not verified |
| Skills | Not verified | Not verified |
| Rules / instructions | Verified | Not verified |
| Hooks | Not verified | Not verified |
| Memory | Not verified | Not verified |
| Bring your own key | Not verified | Not verified |
| Local models | Not verified | Not verified |
| Cloud agents | Verified | Not verified |
| Parallel agents | Not verified | Not verified |
| Check before switching | AI credits and feature access depend on plan and surface. Flex allotments can change. | Confirm compatibility with your IDE and the included AI allowance before subscribing. |
Coding agents
Verified capabilities. “Not verified” is an evidence gap, not a claim that a feature is unavailable. Features can depend on plan, platform and selected agent.
| Decision | Claude Code 2026-09-21 | Codex 2026-09-21 | Grok Build 2026-09-22 | Kimi Code CLI 2026-09-22 | Gemini CLI 2026-09-22 | GitHub Copilot CLI 2026-09-22 | OpenCode 2026-09-22 | Qwen Code 2026-09-22 | goose 2026-09-22 | Aider 2026-09-22 | Amp 2026-09-22 | Factory Droid 2026-09-22 | Kilo Code CLI 2026-09-22 | Cline CLI 2026-09-22 | Cursor CLI 2026-09-22 | Kiro CLI 2026-09-22 | Qoder CLI 2026-09-22 | CodeBuddy Code 2026-09-22 | Mistral Vibe 2026-09-22 | Pi 2026-09-22 | Continue CLI 2026-09-22 | OpenHands CLI 2026-09-22 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Environment | Coding agent | Coding agent | Coding agent | Coding agent | Coding agent | Coding agent | Coding agent | Coding agent | Coding agent | Coding agent | Coding agent | Coding agent | Coding agent | Coding agent | Coding agent | Coding agent | Coding agent | Coding agent | Coding agent | Coding agent | Coding agent | Coding agent |
| Platforms | macOS, Windows, Linux, Web | macOS, Windows, Linux, Web | macOS, Windows, Linux | macOS, Windows, Linux | macOS, Windows, Linux | macOS, Windows, Linux | macOS, Windows, Linux | macOS, Windows, Linux | macOS, Linux | macOS, Windows, Linux | macOS, Windows, Linux | macOS, Windows, Linux | macOS, Windows, Linux | macOS, Windows, Linux | macOS, Windows, Linux | macOS, Windows, Linux | macOS, Windows, Linux | macOS, Windows, Linux | macOS, Windows, Linux | macOS, Windows, Linux | macOS, Windows, Linux | macOS, Linux |
| Good starting point for | Developers who want an agent alongside their preferred editor. | Developers evaluating a coding agent without making it their editor. | Grok-based development and embedding an agent in terminal scripts or ACP clients. | Repository exploration and coding with Kimi, including migration from the older kimi-cli. | Developers using Google accounts, Gemini API keys or Vertex AI. | Developers who combine terminal work with GitHub repositories and code review. | Developers who want to choose providers without replacing their editor. | Developers using Qwen models and Alibaba ModelStudio for coding tasks. | Developers automating local workflows with extensible tools and provider choice. | Small, reviewable code changes while keeping your own editor and model provider. | Teams moving between local work and long-running tasks in hosted Orbs. | Teams taking a local coding workflow into scripted tasks and multi-agent projects. | Developers who want to choose models and switch between planning, debugging and implementation. | Cline users extending their editor workflow into scripts and terminal sessions. | Cursor users who also work over SSH, in terminal editors or in automation. | AWS and Kiro users combining project guidance with terminal development. | Qoder users who prefer a terminal workflow without launching the IDE. | Developers using CodeBuddy accounts or Tencent enterprise authentication. | Developers choosing Mistral models, compatible endpoints or local inference. | Developers building a small, customizable agent workflow around their own tools. | Developers connecting Continue configurations to terminal work and CI scripts. | Maintaining an existing OpenHands CLI setup and understanding its configuration. |
| Pricing model | Claude subscription or provider usage billingSee official pricing | ChatGPT plan or API usagePaid from $8/mo | Account access or metered API usageSee official pricing | Kimi Code account or provider API usageSee official pricing | Account quota or metered Gemini / Vertex AI usageFree access; see limits | Copilot plan and its usage allowanceFree access; see limits | Free CLI; model-provider or optional service chargesSee official pricing | ModelStudio or configured provider billingSee official pricing | Free CLI; model-provider or optional service chargesSee official pricing | Free CLI; model-provider or optional service chargesSee official pricing | Plan allowance, usage credits and separate hosted computeSee official pricing | Factory subscription, rolling limits and optional extra usageSee official pricing | Provider usage or Kilo service billingSee official pricing | Provider usage or Cline account billingSee official pricing | Cursor subscription and usage allowanceSee official pricing | Kiro credits and optional add-on usageFree access; see limits | Qoder plan and Credits usageSee official pricing | Regional CodeBuddy plan and usage allowanceSee official pricing | Hosted API usage or local inference costsSee official pricing | Free CLI; model-provider or optional service chargesSee official pricing | Account configuration or model-provider usageSee official pricing | Legacy client; configured provider and infrastructure costsSee official pricing |
| Free access | Not verified | Available; limits apply | Not verified | Not verified | Available; limits apply | Available; limits apply | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Available; limits apply | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified |
| Model access | Anthropic subscription, Console or supported third-party provider. | Model access varies by ChatGPT plan or API credentials. | Grok models and configurable API endpoints; inspect the active model before running a task. | Kimi Code OAuth, Kimi Platform API keys, or manually configured providers. | Google sign-in, Gemini API key, or Vertex AI credentials. | Select from the models enabled for your Copilot account and organization. | Connect a supported provider and select its models; custom endpoints need their own configuration. | Alibaba ModelStudio API access or Token Plan; other providers depend on configuration. | Choose a supported model provider or an ACP agent as the execution provider. | Use supported hosted APIs or local backends such as Ollama and LM Studio. | Amp routes work across models and supports configured subscriptions or provider keys. | Use Factory’s available model pool or configure custom model credentials. | Connect a supported model provider; available models and prices follow that provider. | Cline Provider, ClinePass or your own supported provider credentials. | Model access follows your Cursor account and the CLI’s current model selector. | Kiro-managed models available to the account; free-tier and regional limits apply. | Model access and Credits depend on your Qoder account and selected mode. | Available models depend on the regional site or enterprise domain used to sign in. | Mistral-hosted models, local models or an OpenAI-compatible endpoint. | Supported provider API keys or account logins; check provider terms and current compatibility. | Continue-configured models or a directly supplied Anthropic API key. | Configure the LLM provider supported by your installed version. |
| Agent workflow | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified |
| CLI surface | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Verified |
| MCP | Verified | Verified | Verified | Verified | Verified | Verified | Verified | Not verified | Verified | Not verified | Verified | Verified | Verified | Verified | Not verified | Verified | Not verified | Not verified | Verified | Not verified | Verified | Verified |
| Skills | Verified | Verified | Verified | Verified | Verified | Not verified | Verified | Not verified | Not verified | Not verified | Verified | Verified | Verified | Not verified | Not verified | Not verified | Not verified | Not verified | Verified | Verified | Not verified | Not verified |
| Rules / instructions | Not verified | Not verified | Verified | Not verified | Verified | Not verified | Verified | Not verified | Not verified | Not verified | Verified | Not verified | Not verified | Not verified | Not verified | Verified | Verified | Not verified | Not verified | Verified | Not verified | Not verified |
| Hooks | Verified | Not verified | Verified | Verified | Verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified |
| Memory | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified |
| Bring your own key | Not verified | Not verified | Verified | Verified | Verified | Not verified | Verified | Not verified | Verified | Verified | Verified | Verified | Verified | Verified | Not verified | Not verified | Not verified | Not verified | Verified | Verified | Verified | Verified |
| Local models | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Verified | Not verified | Not verified | Not verified |
| Cloud agents | Not verified | Verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Verified | Verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified |
| Parallel agents | Not verified | Not verified | Not verified | Verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified |
| Check before switching | Choose between subscription and provider billing deliberately; the editor is a separate choice. | ChatGPT access and API billing are separate options. Features and limits depend on sign-in method. | Browser login and API access have different billing paths; a chat subscription is not an API allowance. | The current Node-based release replaces the Python/uv version; migration does not copy login credentials. | Quotas depend on authentication. Workspace accounts may need a Google Cloud project. | Organization administrators must enable CLI access; allowances follow the Copilot plan. | Provider credentials, subscription eligibility and API costs differ; not every login method is interchangeable. | Current setup uses provider authentication; do not assume an older free OAuth quota still applies. | The client is free; external model providers and connected services may charge separately. | Model quality and context limits vary; inspect automatic Git changes and the selected files. | Agent usage and Orb compute have separate allowances; Windows terminal use requires WSL. | Rolling limits, extra usage and BYOK rules differ; Missions may require extra usage to be enabled. | Current documentation targets CLI 1.0+; older CPUs may need a baseline binary. | Review auto-approval settings before running tasks; headless execution can change files and run commands. | The CLI is a separate working surface from the editor; account limits still apply. | CLI and IDE use the same account allowance; enterprise and regional access can differ. | Native installation is now preferred over legacy npm; Windows ARM64 is not supported in the checked guide. | China, international and enterprise endpoints offer different account and model access. | Local operation still needs a suitable model and hardware; displayed cost estimates are not spending caps. | Extensions can expand its permissions and behavior; the default tools can already edit files and run Bash. | Headless mode excludes tools that require approval unless you explicitly allow them. | The official repository says this CLI is no longer actively maintained and recommends Agent Canvas. |
AI work assistants
Verified capabilities. “Not verified” is an evidence gap, not a claim that a feature is unavailable. Features can depend on plan, platform and selected agent.
| Decision | TraeWork 2026-09-21 |
|---|---|
| Environment | AI work assistant |
| Platforms | macOS, Windows, Web |
| Good starting point for | Task-oriented work where a conversation and deliverable matter more than navigating source files. |
| Pricing model | TRAE membership + metered usagePaid from $20/mo |
| Free access | Available; limits apply |
| Model access | Model access depends on the selected plan, region and current product surface. |
| Agent workflow | Verified |
| CLI surface | Not verified |
| MCP | Not verified |
| Skills | Not verified |
| Rules / instructions | Not verified |
| Hooks | Not verified |
| Memory | Not verified |
| Bring your own key | Not verified |
| Local models | Not verified |
| Cloud agents | Verified |
| Parallel agents | Verified |
| Check before switching | Check whether your task needs a full IDE, debugger or extension before replacing your development environment. |
Each product guide links to its official evidence. Read the verification method . Workflow fit is editorial judgment, not a performance benchmark.
Related TRAE guides
继续阅读
更多围绕相同主题、协议或工具的文章。
引用的工具
浏览与本文主题相关的目录条目。
