Best Augment Code Alternatives in 2026

If you are weighing up Augment Code alternatives, the short answer is this: Cursor is the best general-purpose replacement, Gemini CLI is the best free one, and Claude Code is the strongest pick for developers who work in a terminal. Augment Code has not gone anywhere and remains excellent at one specific job - giving an agent real awareness of a sprawling codebase - but as of 14 October 2026 it has no permanent free plan, meters usage through a shared dollar pool, and has rewritten its pricing twice in twelve months.

Everything below was checked against vendor pages on 14 October 2026. Prices are monthly unless stated.

Augment Code alternatives at a glance

Alternative	Best for	Free plan	Starting price	Main advantage	Main limitation
Cursor	All-round AI coding	Yes	Free; Plus $8/mo	Editor, agent, cloud agents and PR review in one subscription	Third-party model usage billed separately at list price
GitHub Copilot	GitHub-native teams	Yes	Free; Pro $15/mo	Native PR review and predictable per-seat pricing	Free tier is 2,000 completions and 50 chats a month
Claude Code	Terminal-first power users	No	$17/mo annual ($200 up front)	Strongest agentic loop, works locally without a remote index	Not in the free Claude plan; usage-capped
Gemini CLI	Free usage	Yes	Free; 60 req/min, 1,000 req/day	Apache 2.0 with a genuinely usable daily allowance	Tied to the Gemini model family
Cline	Open-source in the IDE	Yes	Free; pay inference or BYOK	No subscription, no seat fee, many providers	You own the model bill and the support burden
Aider	Budget terminal pairing	Yes	Free; pay provider directly	Git-native commits, connects to almost any model	No editor, no PR product, no team controls
Amp	Security-conscious teams	Yes	Hobby free; Individual $20/mo	BYOK with no token fees, strong audit and access controls	Terminal-first, buy model access separately
Kiro	AWS-aligned teams	Yes	Free (50 credits/mo); Pro $20/user/mo	Cheapest credible paid per-seat agentic IDE	Free tier limited to open-weight and Sonnet 4.5
Codex	ChatGPT subscribers	Yes	Included with ChatGPT Free and up	Parallel cloud agents that survive closing the laptop	Usage limits bite on long sessions
Qodo	PR quality gates	No	14-day trial; Pro Team from $30/mo	Reviews every PR across GitHub, GitLab, Bitbucket, Azure DevOps	Reviews only - no code generation
Devin	Autonomous delegation	Yes	Free $0; Pro $20/mo	Whole tickets handed to a cloud agent	Quota pricing; unusual team fee structure

Why look for a Augment Code alternative?

The free tier disappeared, and that is the most common trigger. Augment’s pricing page on 14 October 2026 shows exactly three options: Standard at $20/month flat, Business at $100/month flat, and Enterprise on request. There is no free row. The only zero-cost path is a trial credit pool, and the vendor’s own October 2025 announcement states that trial users receive 30,000 credits “upon signing up with a valid credit card.” A payment method is a meaningful wall when you are trying to evaluate an agent honestly on a side project.

Pricing has moved twice in a year. Augment moved from per-user-message plans to credits on 20 October 2025, then restructured again around flat pooled plans with a $20 Standard tier added in September 2026. The October 2025 post is candid about why: it disclosed a single user on the old $250 Max plan issuing 335 requests an hour, every hour for 30 days, and approaching $15,000 a month in cost. That is a defensible business decision. It is also an admission that your bill is a function of someone else’s behaviour. Usage metering now works as a shared dollar pool across up to 50 seats, with LLM inference billed at provider list price plus a flat 40% service fee, and top-ups valid for 12 months.

There is no open-source or self-hosted option. Every competitor below that offers one - Gemini CLI, Cline, Aider - lets you read the source, fork it, and run it against your own keys or your own hardware.

The surface is narrow. Augment is a VS Code and JetBrains extension plus the Auggie CLI. It does not replace your editor, and it does not give you a code host or a multiplayer agent session. If you want an agent-native editor or a shared team workspace, Cursor or Amp cover ground Augment does not.

Windsurf is gone as a standalone option. If your shortlist still contains Windsurf, that needs revisiting. Cognition announced a definitive agreement to acquire the agentic IDE - product, brand, IP and business - on 14 July 2025, and has since folded it into Devin. Cognition’s own announcement put Windsurf at $82M ARR with 350-plus enterprise customers at the time; the market has moved on.

How I chose these alternatives

Prices, plan limits and free-tier details come from vendor pricing pages and documentation dated 14 October 2026.

Selection criteria, in order of weight: real workflow overlap with Augment Code (an agent that reads and edits a repository, plus review or automation around it); active product with a live pricing page today; breadth of genuine alternatives rather than padding. I scored each on output quality, ease of adoption, pricing clarity, free-tier value, integrations, collaboration, customisation, reliability, privacy and security posture, platform support, migration difficulty and scalability. Ratings on a five-point scale are editorial judgements.

Where a vendor’s pricing page and a third-party tracker disagreed, I used the vendor’s page. For example, several trackers still quote GitHub Copilot Pro at $10; the official plans page on 14 October 2026 lists Pro at $15 with $10 of base AI credits plus a $5 flex allotment.

1. Cursor - Best overall

Cursor is a fork of VS Code with the agent woven through the editor, plus a terminal CLI, cloud agents, hosted code through Origin, and PR review through Bugbot. It is the closest single-product swap for Augment Code because it covers the same surface area without needing a second tool.

Pricing, read on Cursor’s pricing page on 14 October 2026, is monthly: Free $0, Plus $8, Premium $30, Super $100, Ultra $200, Teams from $30 per user, Enterprise custom. Cloud agents, code hosting and Bugbot review start at Premium. The important caveat sits in the page’s own footnote: third-party model usage is not included in those prices and is purchased separately at API list price.

Free plan: yes, permanently, with limited agent and chat requests. Privacy Mode - no training on your data - is available on every tier including free, and Cursor holds AIUC-1, ISO/IEC 27001, ISO/IEC 42001 and SOC 2 Type II.

Versus Augment Code. Cursor wins on breadth and on adoption cost - you keep one subscription for editor and agent, and the $8 Plus tier is a genuine cheap on-ramp. Augment wins on retrieval. Augment’s Context Engine page makes a specific, testable claim: on the same frontier model, Auggie solved tasks at roughly the same rate as Claude Code using 1.65B total tokens against Claude’s 2.35B, and 1.58B cache reads against 2.27B, for a stated $421.34 saving per benchmark run. Cursor does not publish an equivalent comparison.

Verdict. Switch if you want one tool for everything and can live without Augment’s semantic index. Stay if retrieval quality on a huge monorepo is the whole reason you bought in.

2. GitHub Copilot - Best for teams

Copilot is the lowest-friction move for an organisation already paying for GitHub. It runs in VS Code, Visual Studio, JetBrains, Neovim and Azure Data Studio, through GitHub CLI, on GitHub.com and in GitHub Mobile, and it can be handed work that produces a pull request.

Plans, read on GitHub’s official plans page on 14 October 2026: Free $0, Pro $15/month, Pro+ $70/month, Max $200/month. Organisation plans are billed separately per user - Business $19/user/month with 1,900 AI credits included, Enterprise $39/user/month with 3,900 - and credits pool org-wide at $0.01 each.

Free plan: yes. Copilot Free is capped at 2,000 completions and 50 chat requests per month, which is enough to keep an existing setup running but not enough to run an agent. Code review, cloud agent and app modernisation sit behind Pro and above.

The privacy line that matters most in that same page: GitHub does not use Business or Enterprise data to train its models. From 24 April, GitHub may use interaction data from Free, Pro and Pro+ subscribers for training unless they opt out. Individual-plan users must actively opt out; enterprise users do not have to.

Versus Augment Code. Copilot gives you per-seat predictability where Augment gives you a metered pool. It also gives you review and delegation natively on GitHub. You give up the semantic index.

Verdict. Switch if you want a number finance can forecast per head and you are not chasing codebase retrieval. It is the best answer to “we need something everyone already has.”

3. Claude Code - Best for power users

Claude Code is an agentic coding tool that reads your codebase, edits files, runs commands and integrates with your development tooling. Per its documentation, it runs in the terminal, VS Code, JetBrains, a desktop app and the browser from one engine, with shared CLAUDE.md files, settings and MCP servers across all of them.

The gating fact is on Anthropic’s pricing page, read 14 October 2026: Claude Code is not included in the free plan. Pro is $17/month with annual subscription discount ($200 billed up front) or $20/month billed monthly; Max starts at $100/month. A Claude Console account also works.

Free plan: no. Anyone evaluating Claude Code needs a paid Claude subscription or Console billing.

Versus Augment Code. Claude Code reasons over your local repo without requiring a remote code index, which some regulated teams prefer. Its agentic loop is the most refined of anything on this list. But Augment’s whole thesis is retrieval quality at scale, and Claude Code has no equivalent published token-efficiency comparison.

Verdict. Switch if you are terminal-first and want fewer decisions per task. Stay if you have measured Augment’s retrieval saving and it matters to you.

4. Gemini CLI - Best free

Gemini CLI is Google’s open-source terminal agent. It queries and edits large codebases, generates apps from images or PDFs, runs non-interactively in scripts, supports MCP, and ships with Google Search grounding and a 1M-token context window.

The project’s own README states the headline plainly: 60 requests per minute and 1,000 requests per day with a personal Google account, and Apache 2.0 licensing. A Gemini API key path offers 1,000 requests per day on a mix of flash and pro models.

Free plan: yes, permanently, and large. For reference, Google Cloud’s published quotas put Gemini Code Assist Standard at 1,500 agent-mode and CLI requests per user per day.

Versus Augment Code. You lose the managed context engine, the Cosmos agent orchestration platform and the enterprise layer. You gain a free agent that runs at volume, source you can read, and the ability to switch to a paid API key when you outgrow the allowance.

Verdict. Switch if the trial credit pool is the reason you are hesitating, or if you want to evaluate agentic coding at zero cost before spending anything.

5. Cline - Best open-source option in the IDE

Cline is a free, open-source VS Code extension and CLI with a client-side architecture. You bring your own keys - Anthropic, OpenAI, Google, OpenRouter, Bedrock, Vertex, Groq, Cerebras, DeepSeek and others - or buy inference through Cline’s own provider at cost. No subscription, no seat fee, no vendor lock-in.

Its pricing page, read 14 October 2026, confirms the extension is free for individual developers and that the paid Enterprise tier adds JetBrains support, SSO, SLA, centralised billing, RBAC, audit logs and team management.

Free plan: yes for the software. Note the distinction that matters: Cline is free, but running it costs model inference. That is a fundamentally different cost shape from Augment’s pooled dollars, and for most solo developers it is a large improvement.

Versus Augment Code. Cline has no Context Engine, so on a very large repository it will spend more tokens rediscovering structure. In exchange, every prompt stays on your machine and your spend is itemised by your own provider.

Verdict. Switch if privacy or predictable variable cost is the priority. This is the option I would point at a team that has to justify every dollar to a procurement function.

6. Aider - Best budget option

Aider is terminal pair programming that is tightly integrated with git, so every change lands as a reviewable commit. It connects to OpenAI, Anthropic, Gemini, Groq, LM Studio, xAI, Azure, Cohere, DeepSeek, Ollama, OpenRouter, GitHub Copilot, Vertex AI and Bedrock. The licence is Apache 2.0, confirmed by reading the repository’s LICENSE file.

What makes Aider unusually honest is its published benchmark of cost per task. The leaderboard records per-task costs from under a dollar on some open models to well over $100 on frontier models, alongside edit success rates. You can forecast spend before you start.

Free plan: yes for the software, permanently.

Versus Augment Code. No editor, no PR review product, no dashboard, no compliance story. What you keep is git-quality output and total cost control.

Verdict. Switch if you are solo, cost-sensitive, or you want an agent you can drop into CI. Do not switch an enterprise team onto it without building your own controls.

7. Amp - Best for enterprise

Amp is a fast agent for the terminal, spun out of Sourcegraph and now its own product. Its distinguishing feature is bring-your-own-keys with no Amp token fees, so teams can spend subscriptions they already hold. Teams get shared threads, multiplayer sessions, shared billing with pooled credits, and SAML/OIDC SSO. Enterprise adds pooled credits only with no paid seats, SCIM, spend limits, audit logs, network and MCP access controls, IP allowlisting and regional endpoints.

Pricing read on Amp’s pricing page on 14 October 2026: Hobby free, Individual $20/month for 45,000 minutes of agent time, Teams at no extra charge, Enterprise custom.

Free plan: yes, at Hobby level, with pay-as-you-go beyond your own runners.

Versus Augment Code. Both sell an enterprise platform rather than a single extension. Amp’s controls list is at least as strong, and the BYOK model means your model spend is negotiated separately from your agent spend. Augment’s advantage is its Context Engine and its flat pooled plan with no per-seat charge up to 50 seats.

Verdict. Switch if security review is the gate. Keep Augment if retrieval is the reason you are here.

8. Kiro - Best budget enterprise option

Kiro is AWS’s agentic development environment: IDE, CLI, web, mobile and crew surfaces, plus automation in CI/CD. It is also the product AWS points IDE-plugin users toward, since Amazon Q Developer is ending IDE plugin support on 30 April 2027 and Kiro is the named replacement.

Pricing, read on Kiro’s pricing page on 14 October 2026: Free $0 with 50 credits, Pro $20/user/month with 1,000 credits, Pro+ $40 with 2,000, Pro Max $100 with 5,000, Power $200 with 10,000. Add-on credits cost $0.04 each and roll over for 12 months; plan credits do not roll over. Enterprise adds centralised billing, SSO, usage analytics and security controls.

Free plan: yes, permanently, but it reaches only open-weight models and Claude Sonnet 4.5. Paid tiers add Auto, Claude Sonnet 5 and Claude Opus 5. The free tier is unavailable on Enterprise and in GovCloud, where pricing is roughly 20% higher.

Versus Augment Code. At $20 per user, Kiro undercuts most per-seat rivals. Augment’s flat-fee model still wins for very large teams, since a 50-seat Augment Business workspace costs the same $100 whether two people use it or fifty.

Verdict. Switch if you are standardised on AWS identity and billing. Keep Augment if flat pooled pricing across a big team is worth more than per-seat control.

9. Codex - Best for ChatGPT subscribers

Codex is OpenAI’s coding agent, running in ChatGPT, as an IDE extension, in the CLI and via an SDK. Its differentiator is parallel cloud agents with reusable cloud environments that keep working after you close the laptop, plus automatic GitHub pull request review.

Its official pricing documentation is the clearest statement of the model: Codex is bundled into ChatGPT plans rather than sold separately. Free and Go give limited Codex access; Plus is described as powering “a few focused coding sessions each week” with web, CLI, IDE extension, iOS, automatic code review and Slack integration; Pro adds expanded usage. Local messages and cloud chats share one allowance, weekly limits may also apply, and Plus or Pro users who hit the limit can buy credits rather than upgrade. An API key is a separate path billed at API rates, useful for CI.

Free plan: yes, limited, at the ChatGPT Free tier.

Versus Augment Code. You get multi-agent orchestration without a separate agent platform purchase. You give up a unified codebase index and gain usage limits that bite on exactly the long sessions where Augment’s pooled model helps most.

Verdict. Switch if ChatGPT is already a corporate line item. Otherwise the marginal economics rarely justify a second model of thinking.

10. Qodo - Best specialised option

Qodo, formerly CodiumAI, is an AI code review platform covering GitHub, GitLab, Bitbucket and Azure DevOps, with cross-repo review, a rules system that learns from pull request history, a software map for blast-radius analysis, and a CLI toolbox that other agents including Codex, Claude and Kiro can call.

Its pricing page, read 14 October 2026, is explicit: Pro Team starts at $30/month with pooled credits at $0.012 each, designed for teams up to 30 users, with packs of roughly 2,500, 5,000 or 20,000 credits (about 18, 36 or 144 reviews). Enterprise is custom with SSO, audit logs, BYOK and on-prem options. There is a 14-day trial with no card required, and free access for qualified open source projects - but the vendor states plainly that it does not offer a permanent free tier, and reviews pause when the trial ends.

Free plan: no, in the general sense. Worth checking the open source programme if your project qualifies.

Versus Augment Code. Augment’s code review is included on all plans and is one reason its Enterprise tier exists. Qodo is better at review specifically - cross-repo awareness, rules that self-improve, four forge integrations - and cannot generate code at all.

Verdict. Do not switch. Add it. If review is your main reason for buying Augment, this is a bolt-on rather than a replacement.

11. Devin - Best for autonomous delegation

Devin is Cognition’s agent platform for handing off whole pieces of work, now unified with the former Windsurf IDE into a single Devin Desktop surface. Cognition says the merged company grew from 44 to 350 people and from a $73M to a $500M-plus revenue run rate over the year following the acquisition.

Pricing, read on Devin’s pricing page on 14 October 2026: Free $0 with a light agent quota, unlimited inline edits and tab completions; Pro $20/month with frontier model access and cloud agents; Max $200/month; Teams $80/month for the team plan plus $40 per full developer seat; Enterprise custom.

Free plan: yes, at the Free tier.

Versus Augment Code. Devin optimises for work you delegate rather than work you supervise. Augment optimises for an engineer staying in the loop across a large codebase. They are different products that partly overlap.

Verdict. Switch if you want ticket queues handled asynchronously. Keep Augment if you want to stay in the review loop.

Which Augment Code alternative should you choose?

1. Best overall, and the default recommendation: Cursor, if you want one subscription covering editor, agent, cloud agents and PR review.
2. Best free: Gemini CLI, for 1,000 free requests a day and Apache 2.0 source you control.
3. Best for power users: Claude Code, if you live in the terminal and want the tightest agentic loop.
4. Best for an existing GitHub organisation: GitHub Copilot at $19 per user per month for Business.
5. Best for enterprise security review: Amp, with SCIM, audit logs, spend limits and network/MCP controls.
6. Best budget paid per-seat option: Kiro Pro at $20 per user per month, especially on AWS.
7. Best open source in the editor: Cline, free with bring-your-own-keys.
8. Best open source in the terminal: Aider, free, git-native, with published per-task costs.
9. Best if ChatGPT is already funded: Codex, bundled into your existing subscription.
10. Best bolt-on rather than replacement: Qodo, if code review is the real pain point.
11. Best for full delegation: Devin, if the goal is to hand off tickets rather than supervise them.

When you should keep Augment Code

Keep it if all four of these are true of your situation.

You are working in VS Code or JetBrains and prefer not to change editors. Augment’s extension plus Auggie CLI covers both, and the Oct 2026 changelog shows active work in both - sensitive-file approval prompts for .env files, keys and anything excluded by .augmentignore, plus a JetBrains plugin raised to a 4 GB sidecar heap.

Your codebase is genuinely large and multi-repo. The Context Engine’s published claim - semantic indexing across repositories, commit history, docs and tickets, with proof-of-possession access control - is the specific thing Cursor and Copilot have not matched on paper.

You want code review bundled in. Code Review is available on all plans; Enterprise adds analytics, user allowlists, MCP configuration for Jira, Linear and Notion, multi-org support and unlimited seats and repos.

Your procurement requires a written no-training commitment. Every paid plan excludes AI training on your data under the commercial terms, and the vendor publishes SOC 2 and ISO/IEC 42001 attestations.

And if you have already measured the retrieval benefit - token counts per task, review turnaround, cost per merged PR - and the numbers hold up, the switching analysis you are doing now is probably academic.

When switching makes sense

When the trial credit pool is blocking you. You cannot evaluate an agent properly without a card, and your honest test needs a small repo and two weeks. Gemini CLI removes the wall entirely.

When your bill is unpredictable. Pooled usage across 50 seats means one heavy user drains the shared balance. Copilot Business or Business-per-seat Kiro give finance a number per head.

When you need an agent your security team can audit. Cline and Aider let you read the source. Amp gives you SCIM, audit logs and IP allowlisting. Gemini CLI is Apache 2.0.

When you need an editor-native agent. Augment is an add-in. If you want agent, cloud agents, hosted code and review in one product, Cursor is the destination.

When GitHub is where the work already happens. Native PR review, issue delegation and a cloud agent that opens pull requests are Copilot’s advantage, and enterprise seats are already administered.

When your team has moved on from IDE extensions to terminal agents. Claude Code and Amp are terminal-first by design and fit a workflow built around CI and scripted automation.

When Augment’s platform layer is more than you use. Cosmos, Experts and the orchestration platform are enterprise features. If you are paying for a platform you never touch, a smaller tool wins on value.

How to migrate without chaos

Do not rip and replace on day one. Augment’s Context Engine is also sold as an MCP server that plugs into Claude Code, Cursor, Zed and other MCP-compatible agents. Running your existing agent against the Augment context engine is the cheapest possible experiment: if the retrieval win survives a different harness, the Context Engine is the product you are buying and the harness is incidental.

Export the context layer before you leave. Index the repositories you care about, note which connectors are configured (the Oct 2026 changelog shows Augment connecting GitHub, GitLab, Bitbucket, Azure DevOps, Jira, Linear and Notion), and check whether you rely on any Experts you built. Experts are reusable templates with their own environment, capabilities and memory - they do not migrate themselves.

Move rules and instructions into a portable file. Augment uses .augmentignore, .augment settings, hooks and skills. Claude Code uses CLAUDE.md, Codex and the wider agent ecosystem use AGENTS.md, Gemini CLI uses GEMINI.md. Port the durable parts - ignore rules, coding conventions, workflow instructions - into the format your new tool reads, and re-test rather than assuming parity.

Stagger the team. Run one team on the new tool for a full sprint cycle before moving anyone else. Compare on the metrics you already care about: time to first review, cost per merged PR, and rework rate on agent-authored changes.

Handle the billing change deliberately. If you are moving from Augment’s flat pooled plan to per-seat pricing, your cost per developer goes up as headcount goes up. Model it before you sign, and take the free tiers seriously as a way to run a genuine bake-off at zero cost.

Do not leave code review behind. If code review was part of why you bought Augment, either move it to Copilot or add Qodo deliberately. Losing review coverage quietly is the most common way these migrations go wrong.

Final recommendation

Best overall: Cursor. It is the most complete single-product answer to Augment Code in 2026, with a free tier, a cheap $8 entry point, and coverage of editor, terminal agent, cloud agents, code hosting and pull request review.

Best free: Gemini CLI. One thousand requests a day on a personal account, Apache 2.0, no card required. Nothing else on this list comes close for zero-cost evaluation.

Best for power users: Claude Code. The strongest agentic coding loop available, terminal-first, running locally against your own repository.

Best for teams that want predictability: GitHub Copilot Business at $19 per user per month, with enterprise data explicitly excluded from model training.

Best open source and self-hosted path: Cline in the editor, Aider in the terminal, Gemini CLI if you want Google’s models behind it.

Who should keep Augment Code. Teams working in VS Code or JetBrains on large multi-repo codebases, who want retrieval depth plus bundled GitHub-native code review plus a contractual no-training commitment, and who have measured that the retrieval win pays for the metered usage.

Who should switch. Solo developers and anyone blocked by the absence of a free plan; teams where pooled metered usage has made forecasting impossible; anyone who needs open-source, self-hosted or bring-your-own-key economics; and anyone who needs an agent-native editor or a shared multiplayer agent workspace.

Frequently asked questions

What is the best Augment Code alternative overall? Cursor. It covers the same ground - agentic editing, terminal agent, cloud agents, hosted code and pull request review - on one subscription with a free tier, which makes it the lowest-risk single swap.

Is there a free alternative to Augment Code? Yes, several, and the best is Gemini CLI at 60 requests per minute and 1,000 per day on a personal Google account. Cline and Aider are also permanently free as software, though you pay your model provider directly for inference. The caveat on all three is that none of them ships Augment’s managed context engine or enterprise platform.

Does Augment Code have a free plan? No. On 14 October 2026 its pricing page lists only Standard ($20/month), Business ($100/month) and Enterprise (custom). The free entry point is a trial credit pool requiring a valid payment method - a trial, not a permanent free tier.

How much does Augment Code cost per month? $20/month for Standard and $100/month for Business, each a flat fee covering up to 50 seats with no per-seat charge, plus the included usage pool ($20 and $100 respectively). Beyond that you top up at provider list price plus a flat 40% service fee on LLM usage, and top-ups remain valid for 12 months.

Which alternative has the strongest enterprise controls? Amp, with SCIM directory sync, audit logs, spend limits and reporting, network and MCP access controls, and IP allowlisting with regional endpoints. If you are already standardised on GitHub, Copilot Enterprise at $39 per user per month is the alternative answer.

Can I keep Augment’s codebase intelligence and switch agents? Yes, and it is the first thing I would try. Augment sells the Context Engine as a standalone MCP server for Claude Code, Cursor, Zed and other MCP-compatible agents. That hybrid lets you keep retrieval while changing the harness, at a fraction of the cost and risk of a full migration.

What happened to Windsurf? Cognition signed a definitive agreement to acquire Windsurf - product, brand, IP and business - on 14 July 2025, and has since merged it into Devin. There is no standalone Windsurf product to evaluate today, so drop it from your shortlist.