№ 04Learn With Darin · Field Guide
ChatGPT: a practitioner’s field guide.
The shape of the consumer ChatGPT app in 2026: the new desktop app (Chat, Work, and Codex in one window), web, and mobile. What each plan buys you, what the GPT-6 models changed, and the workflows that survive a year of use.
The shape of ChatGPT in 2026
ChatGPT in 2026 looks very different from ChatGPT in 2024. There are several surfaces, a ladder of plans from Free to Enterprise, and a model picker whose options change based on which one you're paying for. Most of the confusion people have isn't about how to use the product. It's about which version of the product they're using.
The surfaces, as of September 2026:
- Web at
chatgpt.com, the canonical surface. Most features land here first. - The ChatGPT desktop app (Mac and Windows): since July 9, 2026 this is one app with three modes, Chat, Work, and Codex, plus a built-in browser for agentic work. It's available on every plan, including Free.
- ChatGPT Classic: the previous Mac desktop app, renamed and still supported for people who just want the lightweight chat client.
- iOS app: native iPhone/iPad client with Lock Screen widgets, Siri Shortcuts, and Action Button support.
- Android app: native client with Quick Settings tile, widgets, and standard share-intent integration.
- Browser and Office add-ons: a ChatGPT extension/sidebar for Chrome, and ChatGPT for Word (September 2026) for drafting and revising inside a document.
ChatGPT Atlas, OpenAI's standalone browser, is gone: it was discontinued on August 9, 2026, and its browsing features moved into the desktop app. Sora is gone too; the Sora app and web experience shut down on April 26, 2026. If you learned ChatGPT from an older guide (including earlier versions of this one), those two are the biggest things to unlearn.
The model lineup
OpenAI shipped two model generations in the four months since this guide was last updated. The picker no longer shows a long list of model names; it shows thinking levels, and which model sits behind each level depends on your plan:
- GPT-5.6 Luna: the fast, cheap tier. Free and Go users get it by default, with a Think button for harder questions (still Luna, not Sol).
- GPT-5.6 Sol: the flagship of the July 9 release, and what powers the Medium, High, and Extra High thinking levels on eligible paid plans. Plus gets Medium and High; Extra High and Pro need a Pro plan.
- GPT-6 Astra (September 3): OpenAI's most capable model. In ChatGPT it appears as GPT-6 Pro on the Pro plans, Business, and Enterprise.
- GPT-6 Sol and GPT-6 Luna (September 22): faster, cheaper models trained with the same methods as Astra. They rolled out first in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu; Free and Go users can try GPT-6 Luna in the desktop app.
The most important thing to internalize: the thinking level you pick is the one you get. Since September 14, Instant on Plus and Pro no longer escalates to a thinking level on its own (ChatGPT can still switch for safety reasons), and typing "think harder" doesn't change the level either. If a question deserves reasoning, choose it in the picker.
The plans (and the limits that bite)
The pricing table is messy because OpenAI raised, lowered, and reshuffled tiers throughout late 2025 and 2026. OpenAI also stopped publishing most exact message counts, so where it doesn't print a number, neither do I.
| Plan | Cost | Default model | Usage | Notable |
|---|---|---|---|---|
| Free | $0 | GPT-5.6 Luna | Unlimited everyday text chat (abuse safeguards apply); uploads capped at 3/day | Think button (Luna), web search, image generation, voice, the new desktop app. Ads in the US. |
| Go | $8/mo | GPT-5.6 Luna | Higher than Free | Low-cost on-ramp with raised limits. Ad-supported, like Free. No ChatGPT Work. |
| Plus | $20/mo | GPT-5.6 Sol | Unlimited everyday chat; thinking limits unpublished | Medium and High thinking, ChatGPT Work, GPT-6 Sol/Luna in Work and Codex, Custom GPTs, advanced voice. Ad-free. |
| Pro $100 | $100/mo | GPT-5.6 Sol + GPT-6 Pro | 5× Plus | Launched April 9, 2026 as the middle tier. Adds Extra High and Pro thinking levels, and GPT-6 Pro (Astra). |
| Pro $200 | $200/mo | GPT-5.6 Sol + GPT-6 Pro | 20× Plus | Same model access as Pro $100 with 4× the headroom. The "I bill it as a tool" tier. |
| Business | Standard: $20/seat annual, $25 monthly Premium: $100 annual, $125 monthly |
GPT-5.6 Sol | Standard ≈ Plus; Premium = 5× Standard, no 5-hour limit | Org admin, shared workspaces, centralized billing. Standard repriced down in April 2026; Premium seats added August 25, 2026, and you can mix the two in one workspace. |
| Enterprise | Sales | GPT-5.6 Sol + GPT-6 Pro | Pooled / credit-based | SSO, SCIM, audit logs, custom retention, org-wide credit pools. The compliance tier. |
Two things changed how the limits feel since spring. First, Free no longer has a daily text-chat cap: in August 2026 OpenAI made everyday text chat on GPT-5.6 Luna unlimited (subject to abuse safeguards), while uploads, image generation, voice, and the agentic features keep their own limits. Second, the agentic features now share one budget: ChatGPT Work, Codex, and (since September 9) Deep Research inside Work and Codex all draw on the same allowance, and you can buy credits for additional usage when it runs out.
The desktop app, Mac vs Windows
The desktop story changed on July 9, 2026. What used to be three installs (the ChatGPT app, the Codex app, and the Atlas browser) is now one ChatGPT desktop app for Mac and Windows, with three modes in the same window:
- Chat: the conversation you already know.
- Work: the agent that turns a goal into finished sheets, slides, docs, and small sites (Part 05).
- Codex: the coding agent, which shows the technical detail Work hides. Work and Codex share the same plugins. See the Codex guide.
The app also carries the browser that used to be Atlas: tabs, downloads, logins, and (since September 18) Chrome extensions you can install and pin, so agentic browsing happens without leaving the app. The old Mac app lives on as ChatGPT Classic. If all you want is a fast chat window on a hotkey, Classic is still a perfectly good install.
Hotkeys and the menu bar
The single most-used feature of the desktop clients is still the global hotkey. The defaults below are the long-standing ones; if yours differs after the move to the new app, check the app's settings:
macOS
- Default global hotkey: ⌥+Space.
- Customizable in Settings → Shortcut.
- Conflicts with Spotlight only if you've remapped Spotlight off ⌘+Space. Most don't, so Option+Space is clean.
- Menu-bar icon for one-click access; closing the main window keeps it alive.
- Trackpad haptics on Magic Trackpad.
Windows
- Default global hotkey: Alt+Space.
- Conflicts with the legacy "system menu" command on focused windows; change it on day one.
- System tray icon; closing the window minimizes to tray.
- Some users report the hotkey gets unbound after Windows updates; rebinding in Settings fixes it.
- Available via Microsoft Store or direct download from chatgpt.com.
Companion Mode
The desktop clients support Companion Mode, a small floating window that pins ChatGPT next to whatever you're working on. It was the headline feature of the old desktop apps and remains the most-cited reason people install one.
Practical notes:
- On Mac, the companion window is a real always-on-top window. On Windows, it's "always-on-top" with caveats: some full-screen apps will steal focus and hide it.
- Screen sharing from the companion window works on both apps. Mac uses ScreenCaptureKit (the modern API); Windows uses GDI capture, which is slower and chokes on hardware-accelerated video.
- Companion Mode is not the same as ChatGPT Work or agentic browsing. The companion is a window; Work is an agent that takes actions and produces files. Don't expect the floating window to go do things for you.
Voice mode on desktop
Both apps support voice mode, and both apps offer Advanced Voice Mode on paid plans. The Mac app gets a slightly nicer experience: the voice indicator integrates with the menu bar, and the audio routes to your system default cleanly. On Windows, voice mode respects the default communication device, which may differ from your default playback. Exactly the same trap as the Claude app.
File handling
Drag-and-drop works on both, with the same caveats Claude has on Windows: OneDrive on-demand files need to be hydrated first; antivirus may flag the desktop app's clipboard-paste behavior. Both apps cap individual uploads at ~512MB and per-message totals at ~2GB on paid tiers; Free is limited to 3 uploads a day.
One Mac-only nicety: quick look-style PDF preview shows up after upload, so you can confirm the right file went in. Windows shows just a filename.
iOS vs Android
Mobile is where ChatGPT's per-platform divergence is sharpest, because Apple and Google give app developers fundamentally different primitives for OS-level integration.
The basics
From inside a chat the two apps are nearly identical. Same model lineup, same Custom GPTs, same Memory, same image and document upload, same Canvas (when available on your tier). The fork shows up in how you get to ChatGPT from elsewhere on the device.
Voice mode
Both platforms have Standard Voice Mode (free) and Advanced Voice Mode (paid). The technical difference matters:
- Standard Voice: speech-to-text, then a text model, then text-to-speech. Slower, less expressive, but available to all users.
- Advanced Voice: directly processes audio. No text round-trip. Picks up tone, hesitation, accent. Can be interrupted naturally.
If you've never used Advanced Voice, the experience is meaningfully different from any other AI voice product. It's the closest "talking to a person" comes today, and most casual ChatGPT users haven't tried it because Standard Voice is the default.
Where the platforms diverge
iOS
- Lock Screen widgets with one-tap voice: the fastest path to Advanced Voice on any platform.
- Action Button (iPhone 15 Pro and up): assign ChatGPT as your Action Button target. Squeeze, talk.
- Siri Shortcuts: "Hey Siri, ask ChatGPT…" Handoff is functional but jankier than native Siri because Apple gates intents.
- StandBy support: while charging on its side, the phone can show a "tap to talk" affordance.
- Standard share sheet integration; selected text to "ChatGPT" target works in any app.
- Siri can be set as the system fallback for queries it can't answer, and that fallback can route to ChatGPT (Apple Intelligence integration).
Android
- Quick Settings tile: the killer Android feature, one tap from any app, even from inside another full-screen app. iOS has no equivalent surface.
- Home Screen widget with New Chat / Voice / Camera buttons (same triple Claude offers).
- Long-press launcher menu shortcuts: New Chat, Voice, Last Conversation.
- Standard share intent. Share to ChatGPT works from any app.
- No lock-screen widget (Android's lock screen is more constrained than iOS 16+).
- Cannot be set as the system-default assistant, but Bixby/Google Assistant can hand off via intent.
Net: if your interaction pattern is "phone is locked, I want to talk to ChatGPT now," iOS wins by a lot. Lock Screen widget + Action Button + StandBy gives multiple zero-step paths. If your pattern is "I'm always inside another app and want a quick interjection," Android wins via Quick Settings.
Camera, vision, and live video
Both apps support live video on Advanced Voice: the camera streams to the model and you can have a real-time conversation about what it sees. This was OpenAI's flagship demo for advanced voice and it's now broadly available on paid tiers. Both platforms work; iOS gets slightly better motion stabilization on iPhone 15 Pro and newer due to camera-side compute.
Background limits
- iOS handles voice mode in the background reliably. Locking the screen with Advanced Voice running keeps the conversation alive.
- Android's aggressive battery management is the same trap as the Claude app: Samsung One UI's "Sleeping apps" and Xiaomi's MIUI background killers will sever long voice sessions. Allowlist ChatGPT in battery optimization settings.
ChatGPT Work (and what happened to Atlas)
ChatGPT Work launched on July 9, 2026 alongside GPT-5.6, and it's the biggest change to ChatGPT this year. You give it a goal rather than a prompt, and it works across your apps and files, sometimes for hours, breaking the job into steps and coming back with a finished artifact: a spreadsheet, a slide deck, a doc, a small web app. It's OpenAI's answer to Claude Cowork, and it effectively succeeds the old Agent Mode / Operator.
What's distinctive:
- Plugins: Work connects to Slack, Gmail, Google Drive, Salesforce, and dozens of other tools through a plugins directory. Codex uses the same plugins.
- Scheduling: a Work task can run on a schedule, not just when you ask.
- Deliverables, not transcripts: the output is a file you can open, edit, and share. The chat is just the log of how it got there.
- Deep Research lives here too: since September 9, Deep Research is available inside Work and Codex and draws on the same allowance.
- Models: GPT-6 Sol and GPT-6 Luna (September 22) are available in Work for Plus and above; GPT-6 Astra is the strongest model OpenAI has shipped for computer-use-heavy tasks.
Work is included in Plus, Pro, Business, Enterprise, and Edu at no extra list price. It is not available on Free or Go.
The end of Atlas
ChatGPT Atlas, the Chromium-based browser OpenAI launched on Mac in October 2025, was discontinued on August 9, 2026. OpenAI folded its ideas into the desktop app's built-in browser (tabs, downloads, account logins, Chrome extensions) and into the ChatGPT extension/sidebar for Chrome. If you still have Atlas installed, export any bookmarks you care about and move to one of those.
Capability matrix
The features that exist in your account, as of September 2026. "Mac app" and "Win app" mean the new ChatGPT desktop app; Atlas is gone, so it no longer gets a column.
| Capability | Web | Mac app | Win app | iOS | Android |
|---|---|---|---|---|---|
| Text chat | ✓ | ✓ | ✓ | ✓ | ✓ |
| File upload | ✓ | ✓ | ✓ | ✓ | ✓ |
| Image input (vision) | ✓ | ✓ | ✓ | ✓ | ✓ |
| Image generation | ✓ | ✓ | ✓ | ✓ | ✓ |
| Custom GPTs (Plus+) | ✓ | ✓ | ✓ | ✓ | ✓ |
| Canvas | ✓ | ✓ | ✓ | partial | partial |
| Memory | ✓ | ✓ | ✓ | ✓ | ✓ |
| Projects | ✓ | ✓ | ✓ | ✓ | ✓ |
| Tasks (scheduled) | ✓ | ✓ | ✓ | ✓ | ✓ |
| Standard Voice | ✓ | ✓ | ✓ | ✓ | ✓ |
| Advanced Voice (Plus+) | ✓ | ✓ | ✓ | ✓ | ✓ |
| Live video on voice | – | – | – | ✓ | ✓ |
| Deep Research (Plus+) | ✓ | ✓ | ✓ | ✓ | ✓ |
| ChatGPT Work (Plus+) | see note | ✓ | ✓ | see note | see note |
| Built-in agentic browser | – | ✓ | ✓ | – | – |
| Code Interpreter | ✓ | ✓ | ✓ | ✓ | ✓ |
| Companion Mode | – | ✓ | ✓ | – | – |
| Screen sharing | – | ✓ | ✓ | – | – |
| Global hotkey | – | ✓ | ✓ | – | – |
| Lock-screen widget | – | – | – | ✓ | – |
| Quick Settings tile | – | – | – | – | ✓ |
| Action Button | – | – | – | ✓ | – |
| Siri Shortcuts | – | – | – | ✓ | – |
"Partial" on Canvas mobile means: you can view and edit a Canvas document, but the keyboard-driven editing flow is meaningfully degraded vs desktop. Don't write a long doc on mobile if you can help it.
On ChatGPT Work: it's built into the desktop app, which is where OpenAI launched it and where it works best, with the built-in browser and your local files as its working surface. Availability on web and mobile has been rolling out unevenly by plan and region, so check your own account rather than trusting a table.
Limits, surfaced honestly
Thinking levels, not model names
The picker in September 2026 is organized around how much the model thinks: Instant, then a Thinking control with Medium, High, and Extra High, then Pro. Which model sits behind each level depends on your plan (Part 01). The practical rules:
- Instant is for everyday questions. On Plus and Pro it no longer escalates to a thinking level by itself (since September 14), so a hard question on Instant gets an Instant-quality answer.
- Asking in the prompt to "think harder" does not change the level. Use the picker.
- Plus gets Medium and High. Extra High and Pro need a Pro plan; Pro runs on GPT-6 Astra (shown as GPT-6 Pro).
- Free and Go stay on GPT-5.6 Luna, with a Think button that makes Luna reason longer.
The upside of this design: the old silent-fallback problem is much less of a factor. You're not quietly dropped to a smaller model mid-conversation the way you could be in 2025. The downside: when you hit a thinking-level limit, you have to notice and pick a lower level yourself. If a conversation suddenly feels worse, check the picker first.
Context window
Context length depends on the model and plan, and OpenAI adjusts it often enough that I'd check the current help-center numbers rather than trust a figure here. What doesn't change: ChatGPT degrades around context-fill the same way Claude does. Earlier turns get summarized aggressively, then forgotten. For big document sets, a Project with knowledge files, or a ChatGPT Work task pointed at the files, beats pasting everything into one chat.
Attachment ceilings
- ~10 files per message; 512MB per file on paid plans. Free is capped at 3 uploads a day; Plus at 80 files every 3 hours.
- PDFs: long documents get truncated or summarized; split very long ones.
- Images: 20MB each, max 10 per turn. Mobile compresses uploads, so for screenshots of tiny text, prefer desktop.
- Code Interpreter file uploads are scoped to the session and disappear when the session ends. Don't rely on persistence.
Deep Research and the shared allowance
Deep Research used to have its own fixed monthly count per plan. OpenAI no longer publishes one, and since September 9, 2026 Deep Research inside ChatGPT Work and Codex draws on the same allowance as the rest of your agentic work. You can buy credits when you run out. (Sora video is no longer part of the picture at all; the Sora app shut down in April 2026.) Busy weeks drain this faster than you'd expect. Track your usage.
Memory is not a database
Memory has a finite budget of accumulated summaries, and new "facts" displace old ones over time. Don't treat it as durable storage. If you need persistence, use Projects with explicit knowledge files.
Best practices, by surface
Pick the thinking level on purpose.
Instant no longer promotes itself to a thinking level on Plus and Pro, and "think harder" in the prompt doesn't do it either. For anything important, choose High (or Extra High / Pro on a Pro plan) in the picker before you send. Leave Instant for the quick stuff; it's fast precisely because it doesn't reason.
Custom GPTs for shape; Projects for state.
A Custom GPT is a reusable persona: you publish it once and reach for it when you need that role. A Project is a reusable workspace, carrying documents, memory, and conversation history. Mixing them up wastes both. Use a Custom GPT when the instructions are the constant; use a Project when the state is the constant.
Treat Memory as scoped, not global.
Same as Claude's memory: turn off global Memory in Settings → Personalization, and let Projects' own state do the remembering. Memory's surprises (a recipe-context recalled in a code review, etc.) outweigh the convenience for any user beyond the most casual.
Advanced Voice is for dialogue, not dictation.
If you want it to type what you say, use the standard mic icon, not Advanced Voice. Advanced Voice's strength is back-and-forth: interruptions, pauses, exploratory thinking. It's a worse transcription tool than the standard one because it's actively conversing with you while you talk.
Use Work when the output is a file.
If what you want at the end is a spreadsheet, a deck, a doc, or a small site, hand the goal to ChatGPT Work instead of chatting your way there. Give it the sources, the audience, and what "done" looks like, then let it run. Keep Chat for thinking out loud; use Work for deliverables.
Batch your Deep Research prompts.
Deep Research now shares an allowance with Work and Codex, so every whim you fire off comes out of the same budget as your real work. Batch a week's worth of "I should look into X" thoughts into one session and triage which ones deserve Deep Research vs a normal chat on High.
On mobile, set up the one-touch voice path you'll actually use.
iOS users: assign the Action Button to ChatGPT Voice, or pin a Lock Screen widget. Android users: drop the Quick Settings tile into your pull-down. Whichever you pick, do it once and rely on it. The friction of "open app, tap mic" disappears the moment one-touch is one swipe away.
Branch instead of restart.
Same advice as for Claude: edit the message at the point you want to fork, rather than starting over. ChatGPT's branching UI is in the message hover menu on web, and a long-press on mobile. Most users don't know it exists. The conversation tree is more valuable than a new chat almost every time.
A note on privacy posture
OpenAI's defaults shifted in 2025 and again in 2026. The current state:
- By default, on Free / Go / Plus, OpenAI uses your conversations to train future models unless you opt out in Settings → Data Controls → "Improve the model for everyone."
- On Pro, Business, and Enterprise, training opt-out is the default. Pro is the cheapest tier with this default.
- Ads: Free and Go users in the US see labeled sponsored placements, matched on the current conversation, past chats, and memory. OpenAI says advertisers never receive chat content and that it doesn't show ads to accounts it predicts are under 18. Plus and above are ad-free.
- Browsing and agent features in the desktop app (the successor to Atlas's "browser memories") are their own consent surface; manage them separately from the training opt-out.
- Voice mode does store transcripts, not audio.
- Custom GPTs published publicly are accessible by their author (even if conversations with them are private to you) for usage analytics.
- Memory is visible to you in Settings → Memory; you can delete individual entries.
If you don't know where a setting lives, start at the new Privacy Center (rolling out in September 2026 to Free, Go, Plus, Pro, and Business). On web it's under your account menu → Help → Privacy Center; on mobile it's in Settings. It doesn't change anything on its own; it explains chat privacy, memory, personalization, data use, connected apps, and account security, and links to the switch for each.
The one control most people miss: "Temporary Chat", available on every surface. Temporary Chats don't persist, don't update Memory, and don't (per OpenAI) feed training. Use them when you're sharing something you don't want logged.
Troubleshooting playbook
The failure modes you'll trip on, in roughly the order they show up.
Check the picker. The most common cause in 2026 is being on Instant for a question that needed a thinking level: Instant no longer escalates on its own on Plus and Pro. Switch to High (or Pro, if your plan has it) and re-ask. If a thinking level is greyed out, you've hit its limit for now; drop a level or wait for the window to reset.
Likely culprits:
- You're signed into the wrong account (personal vs Business vs Edu); check the avatar.
- You're in a Project's filtered view rather than All Chats.
- You opted in to Temporary Chat and the chat genuinely doesn't persist.
- Sync lag between mobile and desktop; give it 30 seconds, force-refresh.
- iOS: check Background App Refresh is on for ChatGPT in Settings.
- Android: allowlist ChatGPT in battery optimization.
- Both: weak network on cellular degrades Advanced Voice fastest. Standard Voice tolerates it better; if your signal is bad, fall back.
Common after Windows updates. Open the app's Settings → Shortcut and rebind. If rebinding fails, restart the app from the tray (close fully, not just minimize).
It was discontinued on August 9, 2026. Install the new ChatGPT desktop app (its built-in browser now handles tabs, logins, downloads, and Chrome extensions) or use the ChatGPT extension for Chrome. Export any bookmarks still stuck in Atlas first.
Custom GPTs don't have memory of past conversations by design. Each chat starts fresh against the GPT's system prompt and knowledge. If you need cross-conversation memory, use a Project (not a Custom GPT). The two get conflated; they're different mechanisms.
Deep Research has a fast-path it sometimes takes for prompts it deems easy. If you wanted depth and got speed, restate the prompt with explicit depth cues: "search across at least 10 sources, compare them, identify disagreements." That re-routes to the slow path (which can take 5-15 minutes). It draws on your allowance either way.
ChatGPT's agentic browsing is conservative around payment flows and sign-ins, as Operator was before it. It pauses for human confirmation at the point money changes hands; if you walked away during that pause, the task can stall or expire. This is by design. It's the same reason Claude's Computer Use refuses certain kinds of actions. Don't fight it; just be present at checkout.
Closing thought
ChatGPT in 2026 is no longer a chatbot. It's a portfolio of products that all happen to share a model and an account. Work is an agent that ships files. Codex is a coding agent. The desktop app is a browser too. Custom GPTs are a marketplace. Projects are a workspace. Tasks are a scheduler. Memory is a profile. Voice mode is a phone call. The chat box at chatgpt.com is one entrypoint among many.
Treat ChatGPT as several products, not one. Pick the surface that fits the task, not the surface you opened it from yesterday. — TWD
The trap is using Plus from the same web tab for everything because that's how you started. Each surface rewards a different working pattern; learning which is which is the difference between paying $20/month for a chat window and paying $20/month for an actual productivity layer.
If anything in here is out of date by the time you read it, OpenAI's ChatGPT release notes are the source of truth. The pace of change is high; pricing, models, and feature flags shift more or less monthly.