Which AI coding tool should you actually pay for? Cursor, Copilot, Claude Code and more — straight comparison
| Name↕ | Origin↕ | Freetier↕ | Pricing↕ | Features↕ | Ide↕ | Agent↕ | Model↕ | Bestfor↕ |
|---|---|---|---|---|---|---|---|---|
| Cursor 🇺🇸 USA | 🇺🇸 USA | Hobby (free) | $20/mo (Pro), $60/mo (Pro+), $200/mo (Ultra), $40/seat (Teams); Enterprise custom. Credit pools roughly track the subscription price (Pro $20, Pro+ $70, Ultra $400 of other-model usage) | Autocomplete (unlimited Tab on paid), Chat, Agent Mode, Terminal, Multi-file Refactoring, Code Review, Cloud agents, credit-based billing, Auto mode for unlimited use | VS Code (forked) | Yes — Advanced | OpenAI, Claude, Gemini frontier models | Power devs who want AI-native IDE |
| GitHub Copilot 🇺🇸 USA | 🇺🇸 USA | Free tier exists; free for students/OSS | $10/mo (Pro, $15 credits), $39/mo (Pro+, $70 credits), $100/mo (Max, $200 credits); Business $19/user, Enterprise $39/user. Switched from premium requests to usage-based GitHub AI Credits on Jun 1, 2026 — completions stay free on all paid plans | Autocomplete (best), Chat, Agent Mode, Code Review, PR Integration, AI Credits metering, Copilot Workspace | VS Code, JetBrains, Neovim, Xcode | Yes — Good | GPT-4o, Codex, Claude | Most accessible; GitHub ecosystem |
| Claude Code 🇺🇸 USA | 🇺🇸 USA | Not on Free plan | $20/mo (Pro, $17 annual), $100/mo (Max 5x), $200/mo (Max 20x); Team Standard $25/seat ($20 annual), Team Premium $125/seat ($100 annual); Enterprise $20/seat + usage; API pay-per-token. No standalone Claude Code price | Agent Mode (best), Terminal-Native, Multi-file, Testing, Debugging, CLI tool, subagents, prompt caching and Batch API discounts for heavy use | Terminal (CLI) + VS Code ext. | Yes — Best-in-class | Claude Opus 5 / Sonnet 5 | Complex reasoning & refactoring |
| Windsurf 🇺🇸 USA | 🇺🇸 USA | Free tier with limited daily quota, unlimited tab autocomplete | $20/mo (Pro — raised from $15 on Mar 19, 2026), $200/mo (Max), $40/user/mo (Teams), Enterprise ~$60/user custom. Credits were retired in favour of daily/weekly quotas | Autocomplete (unlimited on all plans), Chat, Cascade Agent, Terminal, Daily/weekly quota system, SWE-1.5 proprietary model, Cloud sessions for background dev | VS Code (forked — Windsurf IDE) | Yes — Cascade | SWE-1.5, GPT, Claude, Gemini | Context management & flow |
| Replit Agent 🇺🇸 USA | 🇺🇸 USA | Starter (free): daily agent credits, one published app, limited compute | $20/mo (Core — cut from $25 on Feb 25, 2026; includes $20 credits, up to 5 collaborators); $100/mo (Pro, up to 15 builders, pooled credits, one-month rollover); Enterprise custom | Agent Mode, Full-stack Generation, Deploy, Browser IDE, autonomous long builds, unlimited workspaces, private deployments on Pro | Replit (browser) | Yes — Can deploy apps | Custom, Claude, GPT | Rapid prototyping & beginners |
| Lovable 🇸🇪 Sweden | 🇸🇪 Sweden | 5 credits/day (30/mo cap) | $25/mo (Pro, 100 credits) or $21/mo annual; Business $50/mo (or $42 annual); Enterprise custom. Higher Pro tiers ($100–$200/mo) exist for heavy credit burn | Build Agent, Full-stack Web Apps, Supabase Integration, unified credit pool (build + hosting + AI features), $25/mo free Cloud hosting on paid plans, two-way GitHub sync, custom domains | Web app (Lovable editor) | Yes — App generation | GPT, Claude | Full-stack apps w/ Supabase |
| Bolt.new 🇺🇸 USA | 🇺🇸 USA | 1M tokens/mo (300K/day cap), Bolt branding | $25/mo (Pro, 10M tokens, one-month rollover), $30/user/mo (Teams), Enterprise custom; yearly billing saves up to ~28% | Agent Mode, Sandbox Preview, Deploy, Full-stack Generation, WebContainers, Bolt Cloud (DB/auth/hosting/analytics), custom domains on Pro | Web app (bolt.new) | Yes — Can deploy | Claude (default), GPT, Gemini | Quick MVPs & instant preview |
| v0 🇺🇸 USA | 🇺🇸 USA | Free plan with $5 monthly credits | $20/mo (Premium), $30/user/mo (Team, $30 credits each), $100/user/mo (Business), Enterprise custom; token-based billing with purchasable extra credits | UI Generation, React/Tailwind, Vercel Deploy, Agent Mode, Figma import, v0 API, shared team credits | Web app (v0.app) | Yes — UI-focused | v0 model family, GPT, Claude | React/Tailwind components |
| Codex CLI 🇺🇸 USA | 🇺🇸 USA | Yes — free tier for quick coding tasks | Free; Go $8/mo; Plus $20/mo; Pro 5x $100/mo and Pro 20x $200/mo (ChatGPT Pro was split on Apr 9, 2026); API pay-as-you-go. Codex rides on the existing ChatGPT subscription at no extra charge | Agent Mode, Terminal-Native, Sandboxed, File Editing, Cloud tasks, Code Reviews, Codex web app & IDE extension; token-metered since Apr 2026 | Terminal (CLI) + IDE ext. | Yes — Very good | GPT-5.x-Codex | CLI-first devs; OpenAI native |
| Cody 🇺🇸 USA | 🇺🇸 USA | None — Cody Free and Cody Pro discontinued Jul 23, 2025 | Enterprise only, bundled into Sourcegraph Enterprise (custom quote — there is no standalone or list-price plan). Sourcegraph's separate agentic tool for individual devs is Amp | Autocomplete, Chat, Commands, Code Graph, Multi-repo, Code Search, enterprise context engine | VS Code, JetBrains, Visual Studio, Sourcegraph Web | Yes — Commands | Claude Opus, Gemini Pro, GPT-5.x | Large enterprise codebases & multi-repo |
| Continue.dev 🌍 Open Source | 🌍 Open Source | Open source/free | Free self-hosted | Autocomplete, Chat, Any Model, Extensible, Privacy-first | VS Code, JetBrains | Experimental | Any (local or cloud) | Privacy & model flexibility |
| Amazon Q Developer 🇺🇸 USA | 🇺🇸 USA | Perpetual free: 50 agentic requests/mo, 1,000 LOC Java transformation | $19/user/mo (Pro). Java transformation overage $0.003/LOC after 4,000 LOC per subscriber. No separate Enterprise edition | Autocomplete, Chat, Security Scanning, AWS-native, Code Reviews, Java code transformation, IP indemnity on Pro | VS Code, JetBrains, AWS | Yes | Amazon Titan, Bedrock | AWS developers & security |
| Tabnine 🇮🇱 Israel | 🇮🇱 Israel | None (free plan retired; 14-day trial) | $39/user/mo (Code Assistant), $59/user/mo (Agentic Platform), both billed annually; Enterprise custom; self-hosting adds your own GPU compute cost | Autocomplete (privacy), Chat, On-premise/VPC/air-gapped deployment, train on your code, zero code retention, IP protection | VS Code, JetBrains, Vim, Eclipse | Yes (Agentic Platform) | Own fine-tuned models + bring-your-own LLM | Enterprise privacy & security needs |
| OpenClaw 🌍 Open Source | 🌍 Open Source | Open source/free (MIT) | Free self-hosted (pay model API costs) | Autonomous Agent, Task Automation, Browser Control, File Management, Messaging Apps (WhatsApp, Telegram, Slack, Discord, Signal), Local Gateway, Persistent memory, 387k+ GitHub stars (passed React in Mar 2026). Creator joined OpenAI Feb 2026; the project stays open source under independent governance with no single company holding control rights | Local agent (CLI/gateway) | Yes — Autonomous agent | Any (local or cloud: OpenAI, Anthropic, etc.) | Personal AI automation & local privacy |
| Trae 🇨🇳 China | 🇨🇳 China | Free plan: 5,000 autocompletions/mo + daily token quota; 14-day Pro trial with unlimited Claude Sonnet / GPT-4o | Lite $3/mo, Pro $10/mo (after 7-day trial, $20 basic + bonus usage), Pro+ $30/mo, Ultra $100/mo; token-based usage balance since Feb 2026 | Autocomplete, Chat, SOLO autonomous agent ("context engineer" — prompt to shipped app), Cloud tasks (up to 10 concurrent on Pro), browser Cloud IDE, VS Code fork | VS Code (forked) + Cloud IDE | Yes — SOLO | Claude, GPT-4o, DeepSeek and others | Cheapest full AI IDE — half Cursor's price |
| Qwen3.8-Flash 🇨🇳 China | 🇨🇳 China | Qwen Chat + open weights (Flash-Next on Hugging Face) | Alibaba Cloud global rate card: $0.113/M input · $0.382/M output at up to 1M context (with context caching discount); cached input ~$0.02/M. Open-weight Flash-Next hosted at ~$0.15/$0.47 | Multimodal, 262K→1M ctx, strong agentic coding at a fraction of Opus-class cost, built into Qwen Office standard mode; open-source Qwen3.8-Flash-Next = Qwen4 architecture preview (125B MoE, 6B active, Qwen Community 1.0 licence) | API / Qwen Chat / Qwen Office | Yes — Agentic-coding focused | Qwen3.8-Flash | Ultra-cheap agentic coding & long context |
| GLM-5.3-Flash 🇨🇳 China | 🇨🇳 China | MIT open weights on Hugging Face | $0.15/M in · $0.50/M out list (cached input $0.015/M) — the 50%-off launch promo ($0.075/$0.25) ENDED Sep 9, 2026, so the list price is now what you pay. No free API tier for this model alone; GLM Coding Plan from $18/mo | Natively multimodal (image+video input), 320B-A18B MoE, 1M-token ctx, MIT license, hybrid sparse+linear attention; ~10x cheaper than GLM-5.2 with better agentic scores; previously previewed as 'Ox Alpha'; Terminal Bench 2.1 84.3 | API / Z.ai chat / open weights | Yes — Long-horizon agent tasks | GLM-5.3-Flash | Open-weight multimodal coding on a budget |