Cursor is an AI-first editor built on the VS Code codebase, with agents that work autonomously in parallel, multi-line tab completion, and a choice of frontier models plus Cursor's own Composer models. It runs in the editor, terminal, Slack, and GitHub for PR reviews.
Key features
AI-native editor built on VS Code
Autonomous coding agents (local and cloud, parallel)
Multi-line tab completion
Model choice: OpenAI, Anthropic, Gemini, SpaceXAI, Cursor models
Composer in-house agent models
BugBot automated code review
MCP servers, hooks, and skills
Terminal, Slack, and GitHub PR integration
Watch: Cursor tutorial
Cursor AI Tutorial for Beginners — Volo (VoloBuilds)
Video by Volo (VoloBuilds) on YouTube — plays here without leaving this page.
Cursor models, modes & features
No models or features match your search.
Models
Grok 4.7
Model
Cursor's flagship first-party model, described in Cursor docs as built for the hardest, longest-running tasks. Covered by every paid plan's included weekly usage pool for Cursor models.
Access
Included in all plans' Cursor-model pool; Free and Plus include only Cursor first-party models (Grok 4.7, Grok 4.6, Grok 4.5, Composer 2.5)
Context window
256k standard; long-context tier up to 500k
Flagship model for complex, multi-step, long-running coding tasksStandard and Fast speed tiersLong-context tier above 256k tokens, up to a 500k windowToken rates beyond the included pool (official Models & Pricing table): Standard $2.00 input / $6.00 output / $0.50 cache read per 1M tokens; Fast $4.00 / $12.00 / $1.00
Limits: On the Plus plan it runs at fixed settings (Fast off, medium effort, 256k context window) · Long-context tier bills at 2x standard rates (3x in Fast)
First-party Cursor model in the included usage pool; also a required member of the Cursor Router pool.
Access
Included in all plans' Cursor-model pool (Free and Plus are first-party-models only)
Covered by the included weekly Cursor-models poolToken rates beyond the pool (official table): $2.00 input / $6.00 output / $0.50 cache read per 1M tokens; Fast $4.00 / $12.00 / $1.00Required model for Cursor Router — blocking Grok 4.6 disables the router
Limits: Context window not stated on the official pages opened
First-party Cursor model in the included usage pool, with Standard and Fast speed tiers.
Access
Included in all plans' Cursor-model pool (Free and Plus are first-party-models only)
Covered by the included weekly Cursor-models poolToken rates beyond the pool (official table): Standard $2.00 input / $6.00 output / $0.50 cache read per 1M tokens; Fast $4.00 / $18.00 / $1.00
Limits: Context window not stated on the official pages opened
Cursor's own fast, cost-efficient model, described as capable for most tasks and built for interactive coding.
Access
Included in all plans' Cursor-model pool (Free and Plus are first-party-models only)
Fast, cost-efficient interactive codingCovered by the included weekly Cursor-models poolToken rates beyond the pool (official table): $0.50 input / $2.50 output / $0.20 cache read per 1M tokens; Fast tier $3.00 / $15.00 / $0.50
Limits: Context window not stated on the official pages opened · On Plus it runs at fixed settings (Fast off, medium effort)
Third-party Anthropic model routed inside Cursor; not part of the included pool — billed at API list price from prepaid credit on Premium and above.
Access
Premium and above, Teams, Enterprise, legacy Pro/Pro+/Ultra; billed at API list price from prepaid credit
Selectable in the model picker on Premium and above, Teams, Enterprise and legacy Pro/Pro+/Ultra plansOfficial listed token rates in Cursor: $2.00 input / $10.00 output / $0.10 cache read per 1M tokens
Limits: Not available on Free or Plus plans · On Teams/Enterprise, eligible requests also incur the Cursor Token Rate ($0.25 per 1M tokens) · Context window inside Cursor not stated on the official pages opened
Third-party Anthropic model routed inside Cursor; Cursor docs note Claude Opus handles complex, multi-step tasks well. Billed at API list price from prepaid credit.
Access
Premium and above, Teams, Enterprise, legacy Pro/Pro+/Ultra; billed at API list price from prepaid credit
Limits: Not available on Free or Plus plans · On Teams/Enterprise, eligible requests also incur the Cursor Token Rate ($0.25 per 1M tokens) · Context window inside Cursor not stated on the official pages opened
Third-party Anthropic lightweight model routed inside Cursor at API list price; explicitly excluded from the Cursor Token Rate.
Access
Premium and above, Teams, Enterprise, legacy Pro/Pro+/Ultra; billed at API list price from prepaid credit
Official listed token rates in Cursor: $0.10 input / $0.50 output / $0.01 cache read per 1M tokensExempt from the Cursor Token Rate on Teams/Enterprise
Limits: Not available on Free or Plus plans · Context window inside Cursor not stated on the official pages opened
Third-party OpenAI model routed inside Cursor; Cursor docs note GPT-5.6 Sol handles complex, multi-step tasks well. Billed at API list price from prepaid credit.
Access
Premium and above, Teams, Enterprise, legacy Pro/Pro+/Ultra; billed at API list price from prepaid credit
Limits: Not available on Free or Plus plans · On Teams/Enterprise, eligible requests also incur the Cursor Token Rate ($0.25 per 1M tokens) · Context window inside Cursor not stated on the official pages opened
Third-party OpenAI model routed inside Cursor at API list price; explicitly excluded from the Cursor Token Rate.
Access
Premium and above, Teams, Enterprise, legacy Pro/Pro+/Ultra; billed at API list price from prepaid credit
Official listed token rates in Cursor: $0.20 input / $1.20 output / $0.02 cache read per 1M tokensExempt from the Cursor Token Rate on Teams/Enterprise
Limits: Not available on Free or Plus plans · Context window inside Cursor not stated on the official pages opened
Third-party Google Flash model routed inside Cursor at API list price; explicitly excluded from the Cursor Token Rate.
Access
Premium and above, Teams, Enterprise, legacy Pro/Pro+/Ultra; billed at API list price from prepaid credit
Official listed token rates in Cursor: $0.75 input / $3.75 output / $0.075 cache read per 1M tokensExempt from the Cursor Token Rate on Teams/Enterprise
Limits: Not available on Free or Plus plans · Context window inside Cursor not stated on the official pages opened
Auto model selection: Cursor Router picks a model per request across Cursor models and third-party models. Three modes — Cost (classic Auto), Balance and Intelligence.
Access
Auto is not available on the Plus plan; available on Free (Cursor models) and Premium and above / Teams / Enterprise per plan rules
Cost mode bills as a Cursor-model request from the included weekly pool, whatever model serves itBalance and Intelligence modes (Teams/Enterprise with Router on) bill at the routed model's list price; a third-party route also incurs the Cursor Token Rate unless the model is on the excluded listRouter pool includes GPT-5.5, Claude Opus 5, Claude Fable 5 and Grok 4.6; GPT-5.5 and Claude Opus 5 are recommended, Grok 4.6 is requiredSelectable via model id 'auto-smart' with parameter 'optimize_for' through the SDKs
Limits: Individual plans have Cost mode only · Cursor Router launches for Teams and Enterprise; Enterprise starts with Router off until an admin opts in · Routed model identity is hidden by default; team admins can set visibility to Displayed
Cursor's agent experience in the editor, CLI and Agents window: the pool covers Agent in the editor, CLI and Cloud Agents. Free includes Agent and Chat with limited usage using Cursor models.
Access
All plans; feature depth and included usage scale by plan
Agent and Chat available from the Free plan with limited usageBuilt-in subagents (Explore, Bash, Browser) pick their model automatically; custom subagents can set a model in YAML frontmatterUsage draws from the shared weekly Cursor-models pool
Limits: Plus runs Cursor models at fixed settings (Fast off, medium effort) and has no Cloud Agents or Automations · Regional model restrictions set by providers can remove models from the picker
Cursor's AI autocomplete. The free tier includes Tab completions with Cursor models, per the official pricing help page.
Access
All plans (Free with limited usage)
Tab completions included on the Free plan (with limited usage)Custom model execution such as Tab and Apply is one of the items the Cursor Token Rate is stated to cover on Teams/Enterprise third-party requests
Limits: Free plan Tab usage is limited · Detailed Tab model/context specifications were not on the official pages opened
Cloud Agents and Automations are added at the Premium tier; on Teams they run with shared team context. Their usage draws from the same weekly Cursor-models pool.
Access
Premium and above; Teams and Enterprise
Cloud Agents and Automations from Premium upwardShared team context on TeamsUsage covered by the weekly Cursor-models pool
Limits: Not available on Free or Plus plans · Detailed environment/VM specifications were not on the official pages opened
Cursor exposes programmatic surfaces rather than a raw model-completion API in the pages opened: TypeScript and Python SDKs (e.g. Cursor.models.list() returning model ids such as 'auto-smart' for the Router), plus an Admin API for team usage/spend data. Users can also bring their own provider API keys in Cursor Settings > Models (not available on Plus). Models themselves are consumed inside Cursor surfaces (editor, CLI, Agents window, SDK, iOS app).
API pricing: API-style usage pricing is per-model token pricing, separate from subscription: Cursor's own models draw from the plan's included weekly pool and beyond it bill at listed token rates (e.g. Composer 2.5 $0.50 in / $2.50 out per 1M; Grok 4.7 $2.00 / $6.00). Third-party models (Claude, GPT, Gemini, others) are never in the pool: on Premium and above they bill at the provider's API list price from prepaid credit from the first request; on Teams/Enterprise eligible third-party requests add a Cursor Token Rate of $0.25 per 1M tokens (some models excluded; individual plans never pay it). Top-ups / Auto Top-up prepaid credit funds additional usage.
Cursor editor (Agents window)Cursor CLICursor iOS appCursor SDK (TypeScript)Cursor SDK (Python)Origin (Cursor code hosting)Admin API (team usage and spend)Bring Your Own API Key (provider keys in Settings > Models)
What should you use it for?
Everyday interactive coding and autocomplete
Use Composer 2.5 (Cursor's fast, cost-efficient first-party model) with Tab completions; both draw from the included weekly pool, and Composer is described by Cursor as built for interactive coding.
Hardest, longest-running multi-step coding tasks
Use Grok 4.7, Cursor's flagship first-party model; its long-context tier extends to a 500k window, and it is covered by the included pool on paid plans.
Using frontier third-party models (Claude, GPT, Gemini) inside the editor
Requires Premium or above (or a legacy Pro/Pro+/Ultra plan): third-party models are billed at API list price from prepaid credit from the first request; they are not available on Free or Plus.
Automated pull-request code review
Use Bugbot agentic code reviews, added at Premium for individuals and listed as a Teams feature.
Asynchronous or scheduled coding work
Use Cloud Agents and Automations (Premium and above); on Teams they run with shared team context and draw from the same weekly Cursor-models pool.
Balancing model quality and cost automatically
Use Auto / Cursor Router: Cost mode bills from the included pool; on Teams/Enterprise, Balance and Intelligence modes route per request at the routed model's list price.
Pros
Considered the most feature-complete AI code editor in 2026
Agents can build, test, and demo features end to end
VS Code extensions, themes, and keybindings carry over
Model, plan and pricing information on this page was last verified on 22 October 2026 against official sources. Prices and line-ups change — always confirm on the official website before buying.
Plan structure changed: Cursor is now sold via 'The X Subscription'. The task brief's Hobby/Pro/Pro+/Ultra/Teams names are LEGACY plans per the official pricing help page — not available to new customers; they are recorded in plans[] marked LEGACY with their official legacy prices ($20 / $60 / $200; Teams Standard $40, Premium $120 per user/mo). Current plans are Free / Plus $8 / Premium $30 / Super $100 / Ultra $200 / Teams per-seat.
Context windows: only Grok 4.7's window is stated on the official pages opened (256k standard, long-context tier up to 500k; Plus fixes it at 256k). Context windows for Grok 4.6, Grok 4.5, Composer 2.5 and all third-party models inside Cursor are null — the docs' per-model reference page was referenced but its full table was not opened in this pass.
Release dates: no model release dates are published on the official pages opened, so every 'released' field is null rather than guessed.
Composer version history (Composer 1 / 1.5 / 2) not documented on the pages opened; only Composer 2.5 is listed as current.
Feature depth for Tab, Bugbot and Cloud Agents (internal models, environment specs, limits) is only summarised on the pricing/models docs opened; dedicated feature docs pages were not opened, so limits are recorded conservatively.
JetBrains IDE integration and Slack/GitHub integrations were mentioned only in third-party sources and were NOT verified on official Cursor pages in this pass, so they are excluded from integrations[].
Cursor Token Rate column semantics in the model pricing tables: input/output/cache-read rates were cross-checked against the Grok 4.7 rates stated in prose on the Available models page; the extra per-model column in the tables (e.g. cache-write for Claude models) was not labelled in the fetched text, so cache-write rates are not recorded.
How to choose the right Coding & Development tool
The right coding tool depends on where you write code and how much of your codebase it can understand. IDE integration is the first filter — a tool that lives in your editor, sees your project files and respects your stack will help far more than a general chatbot you paste snippets into. Look for accurate completions, clear explanations of suggested changes, and strong refactoring and debugging support for your main languages. Privacy deserves real attention: check whether your code is used for training, and whether a self-hosted or enterprise option exists for sensitive projects. Price against the hours it actually saves — most developers feel the difference within the first week of real use.
Cursor is an AI-first editor built on the VS Code codebase, with agents that work autonomously in parallel, multi-line tab completion, and a choice of frontier models plus Cursor's own Composer models. It runs in the editor, terminal, Slack, and GitHub for PR reviews.
What are the main features of Cursor?
Cursor offers AI-native editor built on VS Code, Autonomous coding agents (local and cloud, parallel), Multi-line tab completion, Model choice: OpenAI, Anthropic, Gemini, SpaceXAI, Cursor models, Composer in-house agent models, BugBot automated code review, MCP servers, hooks, and skills and Terminal, Slack, and GitHub PR integration.
How much does Cursor cost?
Cursor follows a freemium model: a free tier is available, and paid plans unlock higher limits and advanced features. Check its official website for current plan prices.
What are the pros and cons of Cursor?
Strengths of Cursor include Considered the most feature-complete AI code editor in 2026; Agents can build, test, and demo features end to end; VS Code extensions, themes, and keybindings carry over. On the downside, Pricing restructured multiple times, causing confusion; Overage usage billed at API rates; Agent-generated code still needs careful review.
Production-ready headless browser for AI agents, written in Rust — on-demand layout and rendering keeps a full browser runtime at a light footprint, so agents can fetch and extract pages, search the web and automate browser tasks via CLI, CDP or WebDriver
Best for: AI agent developers who need fast, light browser fetching, extraction and automation
Pricing:Open Source
Limitation: Full visual capabilities need the --layout flag
Free, open-source Mac app that puts every Claude Code session — sub-agents included — in your MacBook's notch, each with its own pixel-programmer character, so you can see who is working, on which model, and who is waiting on you, and answer from the notch
Best for: Claude Code users juggling several agent sessions who want to see and answer them from the notch
Pricing:Open Source
Limitation: Mac only — macOS 15 or later, and made for MacBooks with a notch
Describe a mobile game in plain language and MyPip's AI builds the iOS-style interface and clean React Native (Expo) code, with preview, editing and export to a full project ready for device or App Store, plus in-game purchases, playable ads and multiplayer
Best for: Non-coders and small teams who want a playable mobile game from a description
Limitation: Pricing details are shown only as an image on the homepage, not as readable text
Usage guard for Claude Code and Codex: set weekly, daily or per-project limits and it warns, asks or pauses the agent before the week's allowance burns out, from the terminal or a Mac menu-bar app
Best for: Claude Code and Codex users who keep burning their weekly allowance early
Pricing:Freemium
Limitation: Stale readings can make a pause late, and requests already in flight or tools outside hook coverage are not stopped
Phone remote for coding agents already running in tmux: each session appears as a chat, and permission prompts and questions arrive as buttons whose tap is typed into the same terminal pane
Best for: Developers who run coding agents in tmux and want to steer them from their phone
Pricing:Open Source
Limitation: The phone app is Android-only; iPhone users must use the browser hub instead
Native Mac Markdown app where reviewing an agent's spec works like a pull request: comment on the exact passage over MCP, the agent fixes it, and every fix shows as a diff to accept or send back
Best for: Mac users who review agent-written specs and docs and want code-review discipline
Pricing:Freemium
Limitation: Reviewing, editing and MCP are Pro-only — reading is free
Welcome to LLM Tools Hub! Hi! I'm Hubi 👋 I help you find the right AI tool from LLM Tools Hub. Tell me what you need — for example: “free AI tools for image generation”
