Six months ago, the choice was simple: Copilot for autocomplete, Cursor if you wanted chat + editing. Now there are 15+ serious options, pricing is all over the place, and the "best" tool depends heavily on how you work.
Here's the quick version:
The market is chaos right now. Cursor changed pricing, Windsurf got acquired, new players every week. Here's what's actually worth using.
New tools, pricing changes, and what's actually working. Weekly.
Join 1,800+ devs · Unsubscribe anytime
✓ You're in. See you Thursday.
Six months ago, the choice was simple: Copilot for autocomplete, Cursor if you wanted chat + editing. Now there are 15+ serious options, pricing is all over the place, and the "best" tool depends heavily on how you work.
Here's the quick version:
| Tool | Free tier | Pro price | Best for |
|---|---|---|---|
| GitHub Copilot | 2K completions/mo | $10/mo | Autocomplete purists |
| Cursor | Limited | $20/mo + usageCredits for heavy use | Power users, agents |
| Windsurf | Yes | $15/mo | Cursor alternative |
| Claude Code | No | $20/mo (Claude Pro)Terminal-based | Claude devotees |
| Trae | Unlimited | Free | Budget-conscious |
| Cline | Your API key | ~$0.10-0.50/task | Control freaks |
Still the most capable AI code editor. Composer mode lets you make multi-file changes with a single prompt. The agent can run terminal commands, install packages, fix errors autonomously. For complex refactors, nothing else comes close.
The catch: they moved to credit-based pricing in June 2025. The $20/mo Pro plan gets you 500 "fast" requests. Heavy users are hitting limits mid-project and paying $40-60/month.
If you're doing serious refactoring, greenfield projects, or complex debugging—still the best. If you mainly need autocomplete, it's overkill.
The OG. Still the most-used AI coding tool by a mile. The autocomplete is fast and good. They added a chat panel, agent mode for multi-file edits, and recently a free tier (2K completions/month).
The free tier is interesting—it's enough for light usage, and GitHub is clearly trying to get everyone in the door before competitors do.
Best for: devs who want fast autocomplete without switching editors. The VS Code integration is seamless. Agent mode is catching up to Cursor but not there yet.
Used to be Codeium, the free Copilot alternative. OpenAI acquired them in May 2025 for $3B. Now it's Windsurf—their answer to Cursor.
The "Cascade" agent is good. Not quite Cursor's Composer, but close. The big question is what happens now that OpenAI owns it. Will they lock it to OpenAI models? Raise prices? For now, it's $15/mo and a solid option.
ByteDance dropped this a few weeks ago. Full-featured AI code editor, Claude and GPT-4o built in, no rate limits, no subscription. Yes, really.
The obvious question: why is it free? ByteDance has deep pockets and wants market share. They're eating the API costs to get adoption. Same playbook as TikTok—subsidize growth, monetize later.
Some devs are cautious because it's ByteDance. Others don't care and are saving $240/year. Your call.
Claude + GPT-4o No rate limits Use your own keys optionalTerminal-based coding agent from Anthropic. No GUI—you run it in your terminal, it reads your codebase, makes changes, runs commands. Feels like pair programming with a very smart robot.
The 200K context window means it can understand large codebases. The SWE-bench score (72.7%) is competitive with Devin. But it's terminal-only, which is a dealbreaker for some.
VS Code extension that turns any LLM into a coding agent. Creates files, runs commands, installs packages, debugs. You bring your own API key, so you know exactly what each task costs.
The transparency is the killer feature. Watch it think, see every API call, understand why it made each decision. Great for learning how AI agents actually work.
Any model (Claude, GPT, etc.) Full transparency ~$0.10-0.50/taskNot everyone needs multi-file agents. If you mainly want fast, smart tab completions—these are lighter and often cheaper:
Fastest autocomplete on the market. 300K token context means it understands your whole file. Now powers Cursor's tab completion.
Try it →Been around since 2019. Local model option for enterprises who can't send code to external APIs. Privacy-focused.
Try it →The company became Windsurf, but the free autocomplete product still exists. Good for individuals who want basics for $0.
Try it →AWS's entry. Free tier is generous. Best if you're deep in the AWS ecosystem—understands CloudFormation, CDK, etc.
Try it →Built by ex-Microsoft engineers. Indexes your entire codebase, understands your architecture. Serious about security and SOC 2.
Learn more →From the code search people. Good if you already use Sourcegraph—understands your whole codebase across repos.
Learn more →The autonomous software engineer. Give it a task, come back later to a PR. Expensive but companies are using it to multiply eng output.
Learn more →Copilot that understands your org's code. Indexes your repos, learns your patterns. The safe enterprise choice.
Learn more →On a budget? Start with Copilot free tier or Codeium. If you want more, Trae is full-featured and free (if you're OK with ByteDance).
Want the best agent? Cursor. Pay for Pro+ if you're a heavy user. The multi-file editing is unmatched.
Want Cursor but cheaper? Windsurf at $15/mo. Or Trae for free.
Prefer terminal? Claude Code. The 200K context is great for understanding large codebases.
Want full control? Cline + your own API keys. Know exactly what you're paying, see every decision.
Enterprise? Copilot Enterprise or Augment. Devin if you want to experiment with fully autonomous coding.
More from BuiltForAI