AI Chatbot para sa Coding: Ang 2026 Developer’s Guide sa Libreng AI Code Assistants at Chat-Based Programming Tools

Choosing an ai chatbot for coding in 2026 is harder than it should be, mostly because the category stopped being one category. ChatGPT and Claude started as chat-first assistants, then grew into serious coding agents. GitHub Copilot started as autocomplete, then turned into a full coding agent with chat, reviews, and task assignment. Cursor and Windsurf are now editor-first agent systems, not just clever plugins. Google and AWS both pushed harder into developer workflows too, which means the old “just pick Copilot” advice is outdated.

I checked the public pricing and product pages for the tools in this guide on Abril 12, 2026. The biggest shift is not model quality alone. It is how much work the tool can do without forcing you to babysit every step. The useful questions now are simple: Can it read a real repo? Can it run commands and tests? Can it review a pull request? Does the free tier let you finish anything meaningful before the meter hits a wall? And when you move from solo tinkering to team use, does the pricing still make sense?

One quick reality check before we get into rankings: there is no serious walang kinakailangang pag-sign up AI coding assistant on this shortlist. The tools worth using need an account because they have to connect to your IDE, your terminal, your repository, or a cloud sandbox. If your end goal is not just generating code but launching a customer-facing bot on Messenger, Instagram, or your website, it helps to Tingnan ang Aming Mga Tutorial before you confuse code generation with deployment.

Why an AI Chatbot for Coding Means Two Different Products in 2026

Most weak comparisons fail right here. They stack ChatGPT, Claude, Copilot, Cursor, and Windsurf in one list as if they all do the same job. They do not.

The first camp is chat-first coding assistants. That includes ChatGPT and Claude. You go there when you want architectural thinking, debugging help, long explanations, API design tradeoffs, migration plans, or a second brain that can also write code. The chat is the center of the experience, and the coding features grew around it.

The second camp is IDE-first coding agents. That includes GitHub Copilot, Cursor, Windsurf, Gemini Code Assist, and Amazon Q Developer. These tools care less about polished general conversation and more about staying close to the repo, the editor, the terminal, the PR, and the command loop. They are built to reduce the number of times you leave your dev environment.

That distinction matters because the best tool depends on what slows you down now. If your bottleneck is thinking through a messy codebase, rewriting a migration plan, or understanding a weird error chain across services, a chat-first assistant often feels better. If your bottleneck is repetitive edits, small implementation loops, PR review, and fast movement inside one editor, the IDE-first tools usually win.

When I compare these tools for real work, I care about five things more than benchmark screenshots:

• Lalim ng konteksto: Can the assistant work with a repo, not just a pasted snippet?
• Action depth: Can it edit files, run commands, execute tests, and propose fixes?
• Pricing clarity: Do you understand the cost before your team adopts it?
• Free-tier usefulness: Can you finish a real task for free, or only admire the demo?
• Team fit: Does it support reviews, policies, admin controls, and predictable rollout?

If you keep those five criteria in view, the market gets much easier to read.

The Short Answer: Which Coding Chatbot Should You Actually Pick?

If you just want the buying read without the long tour, here it is.

ChatGPT with Codex is the best overall AI chatbot for coding in 2026. It is still the broadest product if you want one subscription that can brainstorm, review code, edit files, run tests, and work in both local tools and cloud sandboxes.

Claude with Claude Code is the best choice for reading large codebases and planning difficult refactors. It is the calmest assistant in this group when the repo is messy, the context is long, and the right answer requires restraint instead of speed.

GitHub Copilot is still the easiest recommendation for teams already living in GitHub and VS Code. The Free tier is real, the paid tiers are easy to model, and the GitHub-native workflow keeps getting deeper.

Cursor is the best AI-first editor for developers who want autonomous feature work inside the editor. It feels like the most complete “agentic IDE” package right now, especially if you want cloud agents and deeper editor-native automation.

Windsurf is the most compelling flow-state alternative. If you want the editor to track what you are doing, infer context aggressively, and stay close to your actual working rhythm, Windsurf is a serious contender.

Gemini Code Assist is the strongest free option for students, hobbyists, and anyone who wants an ai chatbot para sa pag-aaral programming. Google’s free daily limits are unusually generous, and the product is getting more practical in terminal and agent mode.

Amazon Q Developer is the best AWS-native pick. If your day includes IAM, Lambda, ECS, Java modernization, .NET upgrades, or infra-aware refactors, Q makes more sense than general-purpose chatbots do.

The 2026 Comparison Table: Pricing, Free Plans, and Best Fits

This table is built for actual buying decisions, not vendor theater. Prices below come from official public pricing and help pages reviewed on April 12, 2026.

The pattern is easy to miss if you only look at sticker price. ChatGPT and Claude sell breadth. Copilot, Cursor, and Windsurf sell coding workflow depth. Gemini Code Assist and Amazon Q Developer sell ecosystem leverage. Your best pick usually depends on which of those three you value most.

ChatGPT With Codex Is Still the Best All-Around AI Chatbot for Coding

If you force me to recommend one paid tool to the widest range of developers in 2026, I still land on ChatGPT with Codex. The reason is not that it wins every narrow category. It does not. The reason is that it covers the most ground well enough that you can justify one subscription for a lot of different work: debugging, planning, reviewing code, writing migrations, reading docs, generating tests, and delegating background tasks.

OpenAI’s current help pages put ChatGPT Plus at $20 per month, a new Pro $100 tier, the older Pro $200 tier, at ChatGPT Business at $25 per user monthly or $20 per user monthly billed annually with a two-seat minimum (OpenAI Plus; OpenAI Pro; OpenAI Business). That matters because OpenAI quietly became more flexible for small teams again. Two technical founders or a two-person agency can buy into the team layer without swallowing a five-seat minimum.

The more important change is Codex. OpenAI’s current Codex overview says the coding agent is included with Plus, Pro, Business, and Enterprise or Edu, and for a limited time is also included on Free and Go. OpenAI positions Codex as something you can pair with in your terminal, IDE, or Codex app, or delegate to in the cloud where it edits files, runs commands, executes tests, and can even automate code review in GitHub (Using Codex with your ChatGPT plan).

That product shape matters more than model labels now. A lot of older content still talks about ChatGPT coding as if it were mainly a smart answer box. It is not. The useful version of ChatGPT for developers in 2026 is “chat plus agent plus local tooling plus cloud tasks.” That is why it stays ahead as a general recommendation.

Where ChatGPT shines in practice is mixed work. You can ask it to explain a legacy authentication flow, propose a safer refactor, write the patch, run tests, and then help you write the changelog. That full arc still feels more coherent in OpenAI’s ecosystem than it does almost anywhere else. If your day jumps between code, docs, issues, SQL, bash, APIs, and implementation plans, ChatGPT keeps pace well.

The catch is pricing clarity and limits. OpenAI publishes plan prices clearly enough, but not the same kind of tidy per-day usage story that Google or GitHub now publish for some developer tools. For a solo user that is mostly fine. For a team trying to forecast heavy coding-agent usage, Copilot, Cursor, or Amazon Q can sometimes be easier to budget.

My rule for ChatGPT is simple. Choose it when you want the strongest all-around AI chatbot for coding, not when you want the cheapest editor plugin. If you only care about staying inside the IDE with predictable counters, other tools may fit better. If you want the broadest software assistant, this is still the default.

Claude and Claude Code Are the Calmest Choice for Large Codebases and Refactors

Claude’s edge in 2026 is not hype. It is composure. When a repo is ugly, the history is confusing, and the safest answer is not the fastest answer, Claude still feels unusually good. It is strong at reading long files, comparing approaches without rushing, summarizing architecture, and pointing out risks before it starts spraying edits across the repo.

Anthropic’s live pricing page now shows a more complicated but still readable lineup than older guides reflect. The consumer side is Free, Pro at $17 monthly equivalent with annual billing or $20 monthly, at Max from $100. On the team side, the current page shows a Team standard seat at $20 per seat monthly if billed annually or $25 monthly, plus a premium seat at $100 per seat monthly if billed annually or $125 monthly. The same pricing page also makes it clear that Claude Code at Claude Cowork are part of Pro and above, while the Team premium seat is where Claude Code becomes part of the team rollout (Claude pricing).

Anthropic’s product pages also show how much the coding story has changed. Claude is no longer just a writing-first chatbot that happens to know Python. The current product pages position Claude Code as a coding agent for terminal and IDE use, and the pricing page centers current model families such as Sonnet 4.6, Opus 4.6, at Haiku 4.5 in the live plan comparison (Claude Code; Claude pricing).

Where Claude wins is codebase understanding. If I had to hand an assistant a giant migration diff, a messy service directory, a shaky test plan, and a set of competing implementation choices, Claude is the one I would trust first to explain what is actually happening. It also stays strong when the job is not pure implementation but implementation plus reasoning, like “read these five modules, tell me the real source of the bug, and propose the smallest safe change.”

It is also one of the better tools when the code task overlaps with prose. PR summaries, architecture notes, migration docs, implementation plans, and internal explanations still feel cleaner in Claude than in most competitors. That makes it especially useful for senior engineers, tech leads, and staff engineers who spend a lot of time translating technical work for other humans.

The downside is usage economics. Claude is easy to like and easy to underestimate. A single Pro seat is reasonable. A team where several heavy users need premium-level coding access gets expensive quickly. Claude’s value is real. It is just strongest when you know why you are paying for it: careful reasoning, long context, and codebase interpretation, not just cheap autocomplete.

GitHub Copilot Makes the Most Sense If Your Team Already Lives in GitHub

GitHub Copilot is still the most pragmatic purchase if your team spends the day in GitHub, VS Code, pull requests, and issue queues. That has become even more true now that GitHub has pushed harder into agent mode, coding agents, code review, and model choice. Copilot is no longer “the autocomplete one.” It is the “already inside your workflow” option.

GitHub’s current plans page is unusually clear. Copilot Free ay $0 and includes 2,000 completions plus 50 chat or agent-mode requests per month. Copilot Pro ay $10 per month or $100 per year, with unlimited completions, unlimited agent mode and chats using included models, a coding agent, and 300 premium requests. Copilot Pro+ ay $39 per month or $390 per year na may 1,500 premium requests and broader model access. GitHub’s docs page also lists Copilot Business at $19 per granted seat monthly at Copilot Enterprise at $39 per granted seat monthly (GitHub plans; GitHub Docs).

That lineup makes Copilot the easiest paid upgrade for a lot of developers. Ten dollars a month is still a low-friction buy compared with most competitors, and the free tier is good enough to test honestly. GitHub also keeps tightening the loop between AI and the places developers already work: IDE chat, agent mode, coding agent, pull-request review, GitHub.com chat surfaces, and Model Context Protocol support.

Copilot’s biggest strength is not raw intelligence. It is workflow placement. When the AI tool is already in the repo host, in the PR, in the editor, and in the CLI, you waste less time context-switching. For engineering teams, that matters more than people admit. A slightly weaker answer that lands inside the right tool at the right moment can beat a stronger answer in another tab.

Copilot is also easy to recommend to students and new developers because GitHub still offers paid access paths to verified students, teachers, and maintainers of popular open-source projects on the individual side. That keeps it relevant in the free and low-cost conversation even when other tools look flashier.

The main watch-out is the premium-request economy. GitHub is transparent about it, which I appreciate, but you still need to understand it. If your team keeps reaching for the newest premium models or leans hard on code review and agent tasks, the cheap-looking monthly price can hide the real usage pattern. Copilot stays strongest when you want predictable integration more than you want a pure chat-heavy thinking partner.

Cursor Is the Most Complete AI-First Editor for Developers Who Want Autonomous Feature Work

Cursor’s argument is straightforward: stop bolting AI onto an editor and just use an editor built around AI from the start. That pitch keeps working because the product keeps getting deeper. In 2026 Cursor is not just an assistant inside an editor. It is an editor, an agent, a remote execution layer, and a review stack in one system.

Cursor’s live pricing page currently shows Hobby as free, Pro at $20 per month, Pro+ at $60 per month, Ultra at $200 per month, at Teams at $40 per user per month. The Pro tier adds extended Agent limits, access to frontier models, MCPs, skills, hooks, and cloud agents. Pro+ and Ultra scale model usage higher, while Teams adds shared chats, commands, rules, centralized billing, analytics, RBAC, and SSO (Cursor pricing).

That pricing page only tells half the story, though. Cursor’s docs on Background Agents show how far the product has moved into asynchronous execution. Background agents can work in an isolated remote environment, clone your repo, use a separate branch, and let you send follow-ups or take over whenever you want. Cursor’s Bugbot product pushes further into AI-assisted code review and pre-merge bug catching (Background Agents; Bugbot).

That combination is why Cursor wins a lot of serious developer loyalty. It does not just answer questions. It keeps you inside a working loop where the agent can inspect files, make edits, suggest diffs, and then go work elsewhere while you keep moving. If your definition of an ai chatbot for coding is “a tool that should actually help me ship features,” Cursor is near the top of the list.

It is especially good for developers who want the AI to handle more than one file at a time. Repo-wide edits, feature stubs, migrations, cleanup work, and branch-based background tasks are where Cursor feels worth paying for. If you mostly want quick answers and the occasional completion, it can feel like overkill. If you want the assistant to own more of the implementation path, it feels strong.

The tradeoff is obvious. You are buying into a new editor habit, not just adding an extension to the one you already know. Some teams love that. Some never quite standardize around it. Cursor is best when you want the product to shape your workflow, not when you want maximum continuity with a plain VS Code setup.

Windsurf Feels Best When You Want the IDE to Stay in Your Flow State

Windsurf is the other major answer to the “AI-first editor” question, but it feels different from Cursor in day-to-day use. Cursor often feels like a powerful editor with deep agent tooling. Windsurf often feels like a system that wants to stay aware of what you are doing at all times and keep the AI in sync with that momentum.

Windsurf’s current pricing page lists Free at $0, Pro at $20 per month, Max at $200 per month, at Teams at $40 per user per month, with extra usage billed at API price on the paid plans (Windsurf pricing). Its usage docs add an important detail the main pricing grid does not spell out cleanly: the Free plan includes 25 prompt credits at unlimited Windsurf Tab, while premium Cascade usage burns prompt credits at model-specific multipliers (Usage docs).

The real product story is Cascade. Windsurf’s docs describe Cascade as an agentic assistant with Code and Chat modes, tool calling, planning, linter integration, checkpoints and reverts, MCP support, web search, browser tools, rules, memories, workflows, and even multiple parallel Cascades (Cascade docs; Cascade product page). Windsurf also leans hard into “real-time awareness,” meaning it uses your recent edits, terminal activity, and surrounding context to reduce the amount of restating you need to do.

That is why Windsurf appeals so strongly to a certain type of developer. If you hate narrating your environment every time you ask for help, Windsurf is compelling. The product is designed to infer more and ask you to repeat less. When that works, it feels fast in a way that plain chat tools do not.

Windsurf also makes sense if you want access to multiple model families without locking yourself into one vendor. Its docs show support for Windsurf’s own SWE models as well as other major provider models, and even bring-your-own-key paths in some individual plans (Cascade models).

The weakness is pricing clarity. Windsurf is powerful, but it takes more attention to understand how long your credits will last and when extra usage starts mattering. That does not make it bad. It just means it is a better fit for developers who care more about flow and capability than about the cleanest possible billing story.

Gemini Code Assist Is the Best AI Chatbot for Learning Programming for Free

If your real goal is an ai chatbot para sa pag-aaral, not just a pure output machine, Gemini Code Assist deserves a much closer look than it gets in most roundups. Google has turned it into a serious free developer tool, not a token demo.

The official Gemini Code Assist site now says the individual plan is available at no cost with no credit card required. It also spells out unusually high free usage: 6,000 code-related requests per day, 240 chat requests per day, at 1,000 model requests per day shared across Gemini CLI and agent mode. On the paid side, Google lists Gemini Code Assist Standard at $19 per user per month with annual commitment or $22.80 without it, at Enterprise at $45 per user per month with annual commitment or $54 without it (Gemini Code Assist).

That is an excellent free-plan story in 2026. Google is also leaning into the idea that Code Assist is not just a code-completion tool. The current product page positions it as a conversational assistant in the IDE, a terminal assistant through Gemini CLI, and an agent-capable tool that can perform a wide range of software-development tasks. The supporting Google Cloud docs also call out contextual responses, code completions, function generation, unit-test help, debugging support, and source citations in generated answers (Google Cloud overview).

This is why I keep recommending Gemini Code Assist to students, junior developers, self-taught builders, and freelancers who are still ramping up. A strong ai chatbot para sa pag-aaral should do more than spit out solutions. It should let you ask follow-up questions all day without instantly slamming into a paywall. Google’s daily quota structure makes that possible in a way several rivals still do not.

It is also stronger than people think for Android Studio, Firebase, Google Cloud, BigQuery, and database-flavored workflows. If your learning path touches Google’s ecosystem, the product becomes more useful fast. That does not mean it is the best general team tool for every company. It means its free value is unusually high.

The main caution is ecosystem gravity. Once you move past individual use, the best enterprise logic for Gemini Code Assist shows up inside Google Cloud and adjacent tooling. If your team is GitHub-native, AWS-heavy, or editor-first in a non-Google stack, another tool may still fit your daily workflow better.

Amazon Q Developer Is the Smartest Pick for AWS-Heavy Teams and Modernization Work

Amazon Q Developer is easy to underrate if you think of it as “the AWS one.” That is true, but it undersells the product. Q is not trying to be the best general-purpose coding chatbot for every developer on Earth. It is trying to be the best assistant when your code, infra, and operations already live near AWS. In that lane, it is increasingly practical.

AWS’s current pricing page lists two tiers. The Free tier includes 50 agentic requests per month, access in the IDE or CLI, and 1,000 lines of Java upgrade transformation per month. The Pro tier is $19 per user per month and adds higher limits, more transformation capacity, admin dashboards and controls, and IP indemnity (Amazon Q Developer pricing).

AWS’s product and documentation pages also make the supported surfaces clear. Amazon Q Developer works in major IDEs, on the command line, and in the AWS Management Console. AWS documents support for agentic coding, inline suggestions, chat, MCP servers, security scanning, refactoring support, and transformation workflows across IDE environments (Amazon Q Developer; AWS IDE docs).

Where Q becomes the right answer is when code and cloud context should stay together. Think Java modernization, .NET upgrades, IAM policy confusion, Lambda handler cleanup, container changes, infra-aware debugging, or teams that want one assistant touching both application code and AWS-specific implementation detail. General chatbots can help with those tasks. Q is built around them.

It is also worth noting that Amazon keeps pushing the agentic experience harder. AWS announced the newer agentic coding flow in the IDE in 2025, and the current product pages keep emphasizing that Q can implement features, document, review, refactor, and help with software upgrades instead of just answering questions (AWS agentic coding update).

The reason Q does not rank higher overall is simple: its biggest advantage is context, not universality. If AWS is central to your work, Q deserves serious consideration. If AWS barely touches your week, another tool will probably feel broader, cheaper, or more natural.

How to Choose a Free AI Code Assistant Without Wasting a Week

The fastest way to choose badly is to test these tools with toy prompts. “Build a todo app” tells you almost nothing. Every serious tool on this list can fake competence on a clean demo task. You need a tighter process.

1. Pick one task you genuinely need this week. Good examples: fix a flaky test, add one endpoint, refactor a messy component, write a database migration, or explain a legacy auth flow you do not trust.
2. Run the same task through three tools, not one. Use one chat-first tool, one IDE-first tool, and one free-first tool. A practical mix is ChatGPT, Copilot Free, and Gemini Code Assist.
3. Force each tool to do real work. Do not stop at explanation quality. Ask for the patch, ask it to run or propose tests, and ask it to explain the failure if the first answer breaks.
4. Score the result on five simple questions. Did it understand the repo? Did it touch the right files? Did it ask useful clarifying questions? Did the tests pass or get closer? Did it create extra cleanup work?
5. Track when the free wall shows up. This is where the products separate. GitHub is clear with 2,000 completions and 50 chat or agent requests. Google is generous on daily use. Cursor and Windsurf free plans are useful, but their limits arrive differently. OpenAI and Anthropic free plans are good for trying the experience, but they are not the cleanest tools for predictable free-volume planning.
6. Only pay after one real task succeeds twice. A single good answer proves almost nothing. Two useful outcomes on real work is the better signal.

If you want a rough starting order for free trials, use this. Start with Gemini Code Assist if you are learning or want the most generous daily free quota. Start with GitHub Copilot Free if you are already inside GitHub and VS Code. Start with ChatGPT Free if you want the broadest chat-plus-coding taste test. Then pay only when one tool starts saving enough friction to justify the bill.

Where AI Coding Chatbots Still Break in 2026

The tools are much better now. They are not magic. If you use them like a careful engineer, they save time. If you treat them like infallible junior staff who never need review, they will eventually burn time instead.

The first recurring problem is confident wrong edits. The model reads enough context to sound sure, but not enough history to know why a weird pattern exists. This shows up a lot in auth code, billing logic, distributed system retries, and code that quietly encodes business exceptions. A strong assistant can still choose the wrong abstraction layer to patch.

The second problem is dependency drift. AI tools still love suggesting package versions, APIs, and framework patterns that are almost right. “Almost right” in a real codebase is a good way to waste an afternoon. The safer workflow is still the boring one: read the diff, run the tests, and verify the package docs if the assistant touched dependencies or framework behavior.

The third problem is shallow testing. Most assistants can write tests. Fewer consistently write the tests you actually need. They often overfit to the happy path, mirror the implementation too closely, or stop at surface coverage. That means the test suite can get bigger while your confidence barely improves. This is one of the reasons Claude, ChatGPT, and Cursor pull ahead for experienced developers: they tend to be better at turning a bug into a thoughtful test strategy, not just a syntactic test file.

The fourth problem is security and permission context. Tools that do not know your policies can still recommend risky shortcuts around tokens, session handling, secrets, role checks, or webhook validation. That matters a lot if you are working on Messenger, Instagram, payments, healthcare, or internal admin surfaces.

The practical fix is not complicated. Keep the prompts narrower, review the diff before merge, run the tests, and ask the assistant to explain why the change is safe before you accept it. AI coding tools are best used like a fast collaborator whose work still needs engineering judgment.

How Developers Use AI Coding Chatbots to Ship Messenger, Instagram, and Website Bots Faster

The productive pattern is not “let the AI build my whole chatbot and hope for the best.” The productive pattern is “use the assistant for the parts humans hate repeating, then ship the live conversation flow in tooling designed for real channels.”

In practice, coding assistants are useful for things like drafting webhook handlers, mapping event payloads, cleaning lead data before it hits Google Sheets or a CRM, writing validation around form flows, generating test fixtures for routing logic, or turning a support transcript into a cleaner FAQ structure. ChatGPT and Claude are especially strong when the task mixes code and reasoning. Copilot, Cursor, and Windsurf are stronger when you already know roughly what you want changed and need faster implementation inside the repo.

This is also where developers sometimes buy the wrong product. An AI coding assistant can help you write the glue. It does not replace business messaging features like channel permissions, template management, comment automation, human handoff, broadcasts, analytics, or shared inbox control. Once the prototype is moving toward a live rollout, you are comparing deployment software, not just model quality.

That is the point where it makes sense to compare the delivery layer directly. If you are pricing a live Messenger, Instagram, or website chatbot after the prototype stage, Tingnan ang Presyo ng MessengerBot instead of treating another model leaderboard as the answer. The model helps you build faster. The platform is what actually runs the customer conversation.

When a Coding Chatbot Is Not Enough and You Need a Real Delivery Layer

A coding assistant can help you generate a webhook, a payload parser, an FAQ draft, or a retry strategy. It cannot, by itself, give you a production messaging stack. That is why so many teams end up with a clever prototype and no clean operational system around it.

If your project already needs live replies across Facebook Messenger, Instagram, and your website, the missing pieces are usually operational, not intellectual. You need a bot flow that non-developers can manage. You need forms, routing, triggers, analytics, broadcasts, and a clean handoff path when the automation should stop. AI helps with the logic. It does not eliminate the need for the delivery layer.

That is where hybrid setups work well. Use ChatGPT, Claude, Copilot, Cursor, or another coding assistant to write and debug the technical glue. Use a production platform to manage the customer-facing flow. And if the build has already outgrown starter-level automation depth, Upgrade to MessengerBot Pro instead of piling fragile one-off code on top of a workflow that really needs a proper messaging stack.

Final Verdict: The Best AI Chatbot for Coding by Use Case

There is no one perfect winner because the category split is real. There is, however, a clean winner for each common workflow.

If you only want one answer, buy ChatGPT first. If you mainly live inside the editor and want heavier implementation help, look at Cursor o GitHub Copilot. If you are still learning and want a free ai chatbot para sa pag-aaral code instead of just generating it, start with Gemini Code Assist. And if your code assistant is helping you build a real customer-facing bot, remember that the agent and the live delivery layer are two different purchases.

Sources and Pricing Pages Used for This Guide

• OpenAI: What is ChatGPT Plus?
• OpenAI: About ChatGPT Pro plans
• OpenAI: Using Codex with your ChatGPT plan
• Anthropic: Claude pricing
• Anthropic: Claude Code
• GitHub: Copilot plans and pricing
• GitHub Docs: Copilot subscription plans
• Cursor: pricing
• Cursor Docs: Background Agents
• Windsurf: pricing
• Windsurf Docs: Cascade overview
• Google: Gemini Code Assist
• Google Cloud: Gemini Code Assist overview
• AWS: Amazon Q Developer pricing
• AWS: Amazon Q Developer