Cursor Alternatives

Google Antigravity is an agentic development platform for developers who want AI agents to help with software work. Its agent-first interface organizes work around tasks, artifacts, and feedback. Developers can follow what agents produce and steer the result while other work continues.

Key Highlights

• Orchestrate multiple subagents across development environments
• Run long workflows in the background
• Review artifacts and verification outcomes
• Provide feedback at the task level
• Download the development environment

What Makes It Different

The product treats agent work as something to manage across several tasks rather than a single chat reply. Artifacts provide concrete outputs to inspect, while verification outcomes help developers assess the work before accepting it. This is useful when implementation, research, and testing need separate attention.

Features & Capabilities

The agent-first surface supports background workflows while other agents work on research, implementation, or testing. Developers provide guidance across the interface and its artifacts to change direction as the task develops.

Start with a contained repository task. Inspect the proposed files, run relevant checks, and review any commands or environment access before treating the result as finished. An agent completing a task does not establish that the software meets every requirement.

User Ratings and Testimonials

Evaluate it with a representative development task: compare the artifacts, test outcomes, and amount of manual review needed. Keep a record of unexpected changes and repeat the test on a second task.

Pricing & Value

Individual access is $0/month, with basic weekly rate limits and unlimited Tab completions and Command requests. Google AI Pro and Ultra provide higher limits and flexible AI credit pools. Business access is available through Google Cloud, with consumption-based options and eligible enterprise plans. Compare account eligibility and the usage allowance before sustained agent work.

Devin Desktop, formerly Windsurf, is a coding IDE and command center for local and cloud agents. Windsurf became Devin Desktop on June 2, 2026. The classic editor remains available, while the Agent Command Center brings agent work into the main interface.

Key Highlights

• Coding IDE with the classic editor experience retained
• Agent Command Center for organizing agent work
• Existing editor settings, workflows, and extensions remain relevant
• Free inline edits and Tab completions
• Agent quotas and model access depend on the plan
• Existing legacy plans retain their transition terms

What Makes It Different

The rename brings the IDE under the Devin brand alongside Cloud, CLI, and Review. It does not mean that every Windsurf plugin is the same product as Devin Desktop. The transition FAQ distinguishes the desktop IDE from other integration paths and preserves existing plan terms.

Features & Capabilities

Developers can use the editor while managing coding-agent tasks from the command center. The existing Windsurf experience remains accessible rather than being removed by the rename. Organizations should review the updated application name and deployment guidance when managing desktop installations.

Paid plans provide access to more models and larger agent allowances. Usage depends on the model, task size, and reasoning required. Unlimited Tab completions or inline edits do not mean unlimited paid-model agent work. Legacy subscriptions should be assessed under their own retained terms.

User Ratings and Testimonials

Past Windsurf reviews describe earlier releases. For a current evaluation, test the editor and agent workflow on a representative repository and review the proposed code changes. No general productivity percentage is guaranteed.

Pricing & Value

• Free: light agent quota, limited models, unlimited inline edits and Tab completions
• Pro: $20/month
• Max: $200/month, with higher allowances
• Teams: $80/month team fee plus $40/month per full developer seat
• Enterprise: custom terms

Paid usage allowances refresh daily and weekly. Extra usage is available at API pricing; legacy Windsurf plans may differ.

Claude Code brings AI-powered coding assistance directly into your development environment. Developers can work with repository files and commands within supported development environments. It targets professional developers working on projects of various sizes.

Key Highlights

• Direct integration with development environments
• Access to Claude models under the selected plan and usage limits
• Subscription and API access with different usage allowances
• Explore code, edit files, and run development commands
• Built for both individual developers and teams

What Makes It Different

Claude Code brings assistance into the repository workflow. It can inspect project context and propose edits or commands, while the developer controls permissions and reviews the outcome. Plans provide different usage allowances rather than fixed limits on repository size.

Features & Capabilities

Claude Code offers AI-powered code generation, debugging assistance, and code review capabilities. Available usage depends on the account and plan; repository size alone does not determine access. The tool supports everyday coding tasks and intensive development work through different AI model access levels.

User Ratings and Testimonials

Evaluate the tool on a contained repository task. Inspect edits, run tests, and compare the amount of manual review required. Generated code and executed commands need developer oversight.

Pricing & Value

• Pro: $20/month, or $200/year, with Claude Code access and usage limits
• Max 5x: $100/month with higher usage than Pro
• Max 20x: $200/month with a larger usage allowance

Claude Code also supports Anthropic Console access billed by API consumption. Subscription limits depend on the selected model, context, and task. Review file edits and command outcomes before accepting the result.

Codex is OpenAI's coding agent for developers who want help writing, reviewing and maintaining software. It can work with a repository, explain code, make edits and run development tools. Local clients and eligible cloud environments provide different ways to carry out a task.

Key Highlights

• Repository work from written instructions
• Code editing, explanation and review
• Commands and tests in the working environment
• CLI and IDE extension workflows
• Cloud tasks for eligible accounts and workspaces

What Makes It Different

Codex works on a development task with project context rather than only returning an isolated code snippet. It can inspect files and use tools while making changes. Its suggestions and test output help a developer review a result, but generated code still needs appropriate checks.

Features & Capabilities

Start in a repository and describe the behavior you want changed. Codex can examine existing implementation, edit files and run available checks. Project instructions, environment permissions and installed tools shape what it can do. Review the diff and any remaining failures before accepting changes.

The CLI and IDE extension provide local development workflows. Codex Cloud runs tasks in managed environments for eligible accounts, subject to workspace settings. Cloud work can continue while a computer is asleep, whereas local workflows depend on the device running the task.

User Ratings and Testimonials

OpenAI shows coding workflows and examples for Codex. Evaluate it with a representative task and your repository's checks. A successful demonstration does not establish correctness for a different codebase, dependency set or deployment environment.

Pricing & Value

• ChatGPT plans: Codex is included across plans, including Free and Go, with usage limits
• Codex Cloud: requires an eligible paid account and enabled workspace access
• API access: separate billing applies when using an API key

Allowances vary by plan, task complexity, tools and model. Check account eligibility and usage before starting long-running tasks.

Aider is an open-source AI pair programming CLI that runs in your terminal. It helps developers use LLMs to start a new project or work inside an existing codebase, while keeping the workflow close to git and the command line. You install it locally, choose a model provider, and ask for code changes from inside the repo.

Key Highlights

• Terminal-based pair programming for new and existing codebases
• Works with LLM providers shown in the Aider docs, including DeepSeek, Anthropic Claude, and OpenAI
• Local install via aider-install, with API keys supplied at run time
• Adds selected files to the chat, then edits or creates files based on your request
• Git-aware workflow with diffs, auto-commits, and /undo support

What Makes It Different

Aider is not a hosted editor or browser workspace. Its main difference is that it brings the AI coding loop into a normal terminal session, so developers can work from an existing checkout instead of moving the project into a separate app.

The workflow is explicit: choose the files or let Aider infer them, pick the LLM, and review the diffs it produces. That makes it a better fit for developers who already trust git, shell commands, and local project tools.

Features & Capabilities

A typical session starts by installing the CLI, changing into a project directory, and launching Aider with a model plus the matching API key. From there, you ask it to modify selected files, create new ones, or work through a feature request.

Aider can connect to cloud and local models. Its materials highlight codebase mapping, git integration, IDE/editor usage, images and web pages as context, voice-to-code, linting, testing, and copy/paste workflows for web chat.

User Ratings and Testimonials

Aider does not publish an aggregate rating, but its site includes developer testimonials about daily coding and existing-codebase work. The praise is strongest around terminal ergonomics and the feeling of pairing with an LLM inside a real project.

The tradeoff is that Aider expects developer habits. You need to manage API keys, choose a capable model, inspect diffs, and keep sensitive code or secrets in mind when sending context to external model providers.

Pricing & Value

• Aider CLI: $0, open-source terminal tool you install locally
• Model usage: varies by provider, since you bring your own API key or use a local/free model where available

Aider does not list a hosted subscription. It fits developers who want a free CLI layer and can pay model providers directly when using paid APIs.

Trae is an AI-native IDE for developers who want an assistant inside the editor and an autonomous coding agent for larger tasks. It combines a traditional coding workflow with SOLO mode, custom agent teams, and codebase-aware context.

Key Highlights

• SOLO mode can plan development workflows, use tools, execute changes, and generate and deploy code for review
• IDE mode keeps agentic help inside a familiar coding workflow
• Built-in and custom agents can work as specialists or sub-agents for larger tasks
• MCP support lets agents pull in external resources when the task needs them
• CUE predicts next edits with Tab, including multi-line suggestions
• Privacy notes say code stays local, with temporary upload for indexing

What Makes It Different

Trae is strongest when you want a code editor and a more autonomous builder in the same product. The site describes IDE mode for daily coding and SOLO mode for larger tasks.

Its agent framework is also central. You can use built-in agents, create your own with specific tools and logic, or let agents call other agents as sub-agents.

Features & Capabilities

In the editor, Trae reads repository context and can use online searches or shared documents to improve answers. Custom rules tune AI behavior, while CUE handles fast edit prediction with one Tab press.

For agent work, Trae includes a preview tab that lets agents interact with browser elements, read console logs, and debug in real time. The privacy section says codebase files stay local, while indexing may temporarily upload files for embeddings.

User Ratings and Testimonials

Trae's own testimonials praise the interface, VS Code-like switching, fast code generation, and SOLO's ability to handle larger codebase changes. Cautions include requests for more model settings, local model support, and performance that may depend on network quality.

Pricing & Value

The following rates use monthly billing and automatically renew:

• Free: limited usage, Auto mode, 5,000 autocompletions/month, and 2 concurrent cloud tasks
• Pro: $20/month with $20 monthly usage, unlimited autocomplete, and up to 10 concurrent TraeWork tasks
• Pro+: $60/month with $60 usage (3x Pro) and up to 15 concurrent tasks
• Ultra: $200/month with $200 usage (10x Pro), model early access, and up to 20 concurrent tasks

Yearly billing is a separate option. The current public list contains these four plans; account-specific legacy subscriptions can differ.

Augment Code is an AI coding platform for professional engineering teams that want agents to work across the software delivery loop, not just autocomplete code. Cosmos coordinates agents for triage, authoring, review and verification, with shared organization knowledge and controls for larger codebases.

Key Highlights

• Cosmos runs agents for triage, PR authoring, review, risk analysis and testing.
• Context Engine maps code by structure, including call paths and active code.
• Integrates with Slack, GitHub, Jira, CI, the CLI, MCP and native tools.
• Supports laptops, dev VMs, managed cloud and customer cloud deployment.
• Includes SOC 2 Type II, ISO/IEC 42001, SSO, RBAC, audit logs and SIEM.
• Paid plans state that Augment does not train AI on customer code.

What Makes It Different

Augment Code is positioned less like a single coding assistant and more like an organization-level agent system. The homepage describes specialized experts for each stage: Work Dispatcher, PR Author, Pair Review, Deep Code Review, PR Risk Analysis and Tester.

Its main technical differentiator is the Context Engine. Augment says it maps codebase structure and pulls only the slice relevant to the task, instead of sending broad keyword matches to a model. That matters for teams worried about token cost, context quality and large-repo changes.

Features & Capabilities

Teams can use shipped experts, fork them or build their own templates with separate environments, capabilities and memory. Cosmos adds an expert registry, human-in-the-loop escalation, organization knowledge, scheduling, sandboxes, shared file systems and lifecycle triggers.

The enterprise feature set includes zero data retention, CMEK encryption, VPC deployment, single-tenant instances, BYOK for models, data residency controls, replayable runs, on-prem deployment, HIPAA BAA availability and dedicated account support.

User Ratings and Testimonials

Augment's homepage includes vendor-reported customer outcomes. Those figures are examples of particular deployments, not guaranteed results for every team.

The product suits teams that need shared agent workflows, policy controls and codebase context. Standard and Business support is community-based with tickets through the support portal, while Enterprise adds dedicated support.

Pricing & Value

• Standard: $20/month, flat price for up to 50 seats, with $20/month of included usage.
• Business: $100/month, flat price for up to 50 seats, with $100/month of included usage across LLM, Context Engine and compute.
• Enterprise: Custom pricing for high-volume organizations, unlimited users, custom usage, multi-region compute, SSO, OIDC, SCIM, CMEK, ISO 42001 compliance and dedicated support.

Standard and Business usage is pooled across the team. Extra usage is handled with pay-as-you-go top-ups, and the pricing FAQ lists a 40% service fee on LLM usage, with compute billed separately and no service fee on compute.

Amp is a coding agent and development environment for web, terminal, and native app workflows. It can inspect projects, edit code, and use tools. Its current experience includes orbs: machines created for agent threads with the project's code and tools ready to use.

Key Highlights

• Coding sessions through the web and CLI
• Native macOS, iPhone, and iPad apps
• Orbs that let agents continue working remotely
• Saved and shared threads across clients
• Plugins, skills, MCP, and AGENTS.md instructions
• Model access through supported subscriptions or your own keys

What Makes It Different

Amp combines the coding agent with a development environment. An orb supplies a machine for a thread, so a remote task can continue after you close your laptop. Local CLI workflows and your own runners provide other ways to work.

Features & Capabilities

Create a project from a Git repository, then give the agent a task. The web experience starts threads in orbs, while the CLI can work on your machine or from scripts. Native apps add notifications and ways to follow up on agent work.

Threads and workspace features support collaboration. Plugins and skills customize the workflow, and model routing can use supported external accounts. A consumer subscription must be connected and eligible; it does not automatically cover every model or infrastructure charge.

User Ratings and Testimonials

Amp publishes product examples and vendor-selected feedback rather than a universal independent rating. Review generated code and test changes in your project. Orb resource needs, provider access, and the selected workflow affect cost.

Pricing & Value

Hobby is free, with pay-as-you-go orbs or your own runners. Paid individual tiers start at $20/month. Teams have no additional team charge, and members can use different tiers. Enterprise pricing is custom. Amp lists no token fee for supported external subscriptions or BYOK; provider charges and orb usage still depend on the selected setup.

Kiro is an agentic coding environment for developers who want AI help with software delivery, not just code snippets. It turns a prompt into requirements, design notes, implementation tasks, code, docs, and tests across an IDE, CLI, and web.

Key Highlights

• Spec-driven development turns prompts into requirements, acceptance criteria, designs, and task lists
• IDE, CLI, and web agents for local work, terminal workflows, and cloud sessions
• Agent hooks run background tasks such as docs, tests, or performance checks
• Native MCP support connects Kiro to docs, databases, APIs, and other tools
• Built on Code OSS, with VS Code settings, themes, and Open VSX plugin import
• Per-prompt credit usage shows how much each request costs

What Makes It Different

Kiro is built around specs rather than chat alone. A natural language request becomes structured requirements, an architectural design, and sequenced tasks that map back to those requirements. That gives the agent a documented plan before it starts changing files.

The workflow also extends beyond the desktop editor. Kiro CLI brings agents into the terminal, while Kiro Web runs sessions in isolated cloud sandboxes against GitHub or GitLab repositories. Steering files carry project rules across the IDE, CLI, and web.

Features & Capabilities

In the IDE, developers can review live code diffs, approve changes, diagnose syntax and type errors, generate commit messages, and use image inputs for UI or architecture context. Autopilot mode can reduce step-by-step prompting while scripts and commands still stay under user control.

Kiro supports common languages including Python, JavaScript, TypeScript, Go, Rust, SQL, YAML, and HCL. The CLI installs on macOS, Linux, and Windows, and subscriptions also work in ACP compatible IDEs and development automation.

User Ratings and Testimonials

Kiro does not publish a public average rating. Customer quotes praise the spec-driven workflow, background hooks, Terraform and Python support, and stronger implementation plans. Practical limits are visible too: usage is credit-metered, premium model access depends on plan and country or region, and autonomous mode is only in Kiro Web today.

Pricing & Value

• Kiro Free: $0/month, 50 credits, open weight models and Claude Sonnet under limits
• Kiro Pro: $20/month, 1,000 credits, premium models, and optional add-on credit packs at $0.04/credit
• Kiro Pro+: $40/month, 2,000 credits, premium models, and optional add-on credit packs at $0.04/credit
• Kiro Pro Max: $100/month, 5,000 credits, premium models, and optional add-on credit packs at $0.04/credit
• Kiro Power: $200/month, 10,000 credits, premium models, and optional add-on credit packs at $0.04/credit

Team plans add centralized billing, usage analytics, SAML/SCIM SSO through AWS IAM Identity Center, organization management, and enterprise security controls.

Cline is an open-source AI coding agent that runs inside VS Code, your terminal, CI pipelines, or your own product through its SDK. It is built for developers who want an autonomous agent that edits across a whole project, runs commands, and works with supported model providers while you keep control of cost.

Key Highlights

• Coordinated multi-file edits with diffs, checkpoints, and one-click undo on every step
• Plan mode to agree a strategy first, then Act mode to execute with step approval or auto-approve
• Runs bash commands in your terminal and reacts to output live, including servers and tests
• Works with Claude, GPT, Gemini, local Ollama or LM Studio, and any OpenAI-compatible endpoint
• Extends through MCP servers and an SDK, plus Slack, Discord, Linear, and CI integrations
• Apache 2.0 open-source coding project

What Makes It Different

Cline is open source (Apache 2.0) and model-agnostic, with several model-access options, including usage billing, BYOK, and optional ClinePass. Its Plan-and-Act split is the core mechanism: Plan mode lists which files it will touch and the steps it will take, then Act mode carries them out one approval at a time, keeping the agent transparent and reviewable rather than a black box.

Features & Capabilities

You point Cline at a task and it reads the codebase, proposes a plan, then makes coordinated changes across many files while keeping imports, types, and behavior consistent. It runs tests, starts dev servers, and reacts to terminal output live. A .clinerules file teaches it your standards, architecture, and deployment conventions.

Beyond the editor, the CLI runs in scripts, cron jobs, and CI pipelines. You can register custom tools and MCP servers and set up multi-agent teams where a coordinator delegates to specialists. Cline reports contributions from more than 250 contributors.

User Ratings and Testimonials

Evaluate the Plan and Act workflow with representative changes, and inspect commands before approving them. Large tasks can consume substantial model usage. Cline reports more than 11 million installs across supported platforms; an install count is not a count of unique developers.

Pricing & Value

• Open Source: Free for individual developers. Includes the VS Code extension, CLI, MCP marketplace, and multi-root workspaces. Model access uses provider keys, local runtimes, or Cline usage billing; no mandatory subscription or seat fee for the open-source version. Optional ClinePass costs $9.99/month.
• Enterprise: Custom pricing. Adds SSO, role-based access control, model and tool controls, remote configuration, observability, and team management. The JetBrains plugin is also listed among the regular applications.

The extension and model access have separate economics. Compare provider usage charges and optional ClinePass with the tasks you expect to run.

GitHub Copilot is an AI coding assistant that works directly in your editor, suggesting whole lines or entire functions as you type. It is built for developers and teams who want faster everyday coding, code-aware chat, and an agent that can complete multi-file tasks. It runs across VS Code, Visual Studio, JetBrains IDEs, Neovim, and the command line.

Key Highlights

• Inline code completion that understands your project context
• Copilot Chat for explanations, refactors, and test generation
• Agent mode that plans and edits across multiple files
• Choice of frontier models (GPT, Claude, and Gemini families)
• Pull request summaries and AI code review
• Works in major IDEs, GitHub.com, and the CLI

What Makes It Different

Copilot's advantage is its tight integration with the GitHub platform millions of developers already use: completions, chat, code review, and agents all live where the code and pull requests do. The model picker lets teams choose the best model per task instead of being locked to one vendor.

Features & Capabilities

Day to day, Copilot autocompletes code, answers questions about your repository, writes tests, and refactors selections. Agent mode goes further, taking a natural-language task and implementing it across files, then proposing a diff you review.

On GitHub itself it summarizes pull requests, suggests review comments, and helps triage issues, extending AI assistance beyond the editor into the whole workflow.

User Ratings and Testimonials

Evaluate suggestions and agent changes on your own repository, including correctness, review effort, and credit use. Editor support, available features, and models depend on the plan and environment.

Pricing & Value

• Free: limited completions and chat per month
• Pro: around $10/month for unlimited completions and agent access
• Pro+: $39/month with a larger AI credit allowance
• Max: $100/month with a larger allowance for heavier workloads
• Business / Enterprise: from $19-$39/user per month with admin controls

Chat, agents, review, and CLI use consume AI credits. Compare expected usage and spending controls alongside the subscription price.

Qoder is an AI-powered coding assistant that helps developers build and deploy applications faster through intelligent code generation. This tool excels at understanding entire codebases and breaking down complex coding tasks into manageable parts. It targets developers who want to speed up their workflow with AI assistance.

Key Highlights

• Generates and edits code across multiple files; changes require review and testing
• Understands complete project architecture and codebases
• Auto-documentation feature saves time and keeps teams aligned
• Smooth and intuitive interface for coding tasks
• Breaks complex problems into simpler, actionable steps
• Helps developers learn new codebases quickly

What Makes It Different

Qoder stands out by focusing on understanding entire project structures rather than just individual code snippets. Its ability to generate code changes across multiple files sets it apart from basic code completion tools. The auto-documentation feature helps maintain team alignment without extra effort.

Features & Capabilities

Qoder offers comprehensive codebase analysis and intelligent code generation. The tool can write code across multiple programming languages and file types while maintaining project consistency. It provides helpful guidance for complex coding tasks and automatically generates documentation. The AI assistant can quickly analyze existing codebases to help developers get up to speed on new projects.

User Ratings and Testimonials

Qoder demonstrates repository-aware coding and multi-file changes. Assess it using your own codebase and development checks. Inspect generated edits, review dependencies and run relevant tests before accepting a result. A working example in one repository does not guarantee correctness in another.

Pricing & Value

Qoder offers a Free plan and credit-limited paid plans: Pro at $20/month, Pro+ at $60/month and Ultra at $200/month. Eligible new users get a 14-day Pro trial with 300 credits.

The main value of Qoder is its ability to help developers build and deploy applications quickly with AI-powered code generation.