Claude Stopped Waiting for You to Open It

GENERAL

Google Gemini is everywhere, but the reviews tell a different story.

Google Gemini has more than 900 million monthly users. That number sounds like a verdict. It's not — it's a distribution stat. Gemini is baked into Android, Google Search, and Workspace. Most of those users didn't choose Gemini. They just opened their phone. And the branding doesn't help: "Gemini" now refers to a chatbot, a mobile app, a voice assistant, an API, and AI features inside Gmail and Docs. One developer put it bluntly — the name is so overloaded it barely means anything.

What does work is the Google integration. If your day runs through Workspace, Gemini offers real advantages:

+ It can draft inside a Doc, summarize full email threads without copy-pasting, and write spreadsheet formulas in plain English.

+ Deep Research runs multi-step investigations and returns cited reports — legitimately useful for competitive analysis or long-form research.

+ The one-million-token context window lets you upload a 500-page document and query it without splitting it into pieces.

What doesn't work is almost everything else. CNET's review found Gemini got basic phone specs wrong, glitched twice on a recipe query, then told the reviewer "I'm not programmed to assist with that." Writing quality, in our experience, consistently reads more generic than Claude. And the reliability problems go deeper than bad outputs — paying users on Reddit and developers on Google's own forum have documented everything from silent model downgrades to infinite loading loops, calling it one of the most unstable frontier models on the market.

Google recently rebranded its paid tiers into AI Plus ($7.99/month), AI Pro ($19.99/month), and AI Ultra ($99.99+). That tiering matters for the bottom line here.

If you already live inside Google Workspace, the free tier is worth using and the $7.99 Plus plan is a reasonable upgrade. Skip Pro at $19.99 unless you specifically need Deep Research or massive file uploads. For general-purpose AI chat and writing at that price point, Claude Pro, in our experience, is more consistent and more reliable. Gemini's more than 900 million users prove Google can ship to everyone. The reviews prove that shipping isn't the same as being ready.

GENERAL

Claude Pro wins on quality. Its usage limits win on frustration.

After aggregating dozens of reviews, user threads, and long-term switching stories, we found a consistent pattern: writing quality is the main reason people switch to Claude and stay with it. YouTuber Parker Prompts spent 90 days using Claude as his primary tool after canceling ChatGPT Plus. He used to spend 15 to 20 minutes reworking every ChatGPT response before posting it. With Claude, he swaps maybe a line or two. The custom styles feature lets you paste samples of your own writing so Claude can match your tone across conversations — something ChatGPT tends to follow for one reply before reverting to its default headers-and-bullets format.

The broader consensus across user reports is that Claude feels more collaborative where ChatGPT feels more aggressively helpful. People describe switching between the two like switching operating systems — different philosophies of how an AI should talk to you.

Now the catch. Claude Pro costs $20 a month and gives you a usage allowance governed by a rolling five-hour window. Use Sonnet, the default model, and you can send dozens of messages per window. Switch to Opus, the most powerful model, and you might burn through your entire allowance in a handful of prompts. Parker said he once sent a single complex message with a large file and hit the limit before he could send a second. Then you're blocked for up to several hours — not downgraded, blocked.

Anthropic recently doubled its usage limits after securing more compute, which helps. But the fundamental tension remains. Your options:

+ Model management: Keep Sonnet as your default and save Opus for complex reasoning or big documents. It works, but as Parker put it: "On a $20 a month subscription, I shouldn't have to think about which model I'm using before every single message."

+ Upgrade to Max: The $100 or $200/month Max plan gives you 5x or 20x the usage — but even Max still has limits rather than true unlimited access.

Two other gaps worth knowing:

+ No image generation. Claude doesn't have it, and Anthropic hasn't announced plans to add it. If you need visuals, you're opening a second tool every time.

+ Memory works differently. ChatGPT persists information across conversations. Claude's memory focuses on selected user details and preferences — how each feels depends on your setup.

In our view, Claude Pro is an excellent AI subscription for professionals who write, think, or collaborate with text for a living. Many users report a noticeable quality gap in Claude's favor over ChatGPT for writing tasks. But you need to know going in that you're paying for a metered product, and the best model eats through your meter fast. If you mostly use Sonnet and save Opus for the moments that matter, many users find the $20/month Pro plan worthwhile. If you want significantly higher access to the most powerful model throughout the day, you're looking at the $100/month Max plan, though even Max plans still have usage limits rather than true unlimited access.

PRODUCTIVITY

Claude Cowork started as a desktop app in January. As of this week, it runs on web and mobile too — meaning you can kick off a task from your laptop, close the lid, and check results from your phone over lunch. Anthropic also recently launched Claude Tag, a beta feature that drops an always-on Claude directly into your Slack channels. The pattern is clear: Anthropic doesn't want you to visit Claude anymore. It wants Claude to already be wherever you're working.

The surprise is who's actually using it. Anthropic released data from 1.2 million sessions across more than 600,000 users, and less than 9% of usage was software development. The biggest category was business process work — pulling scattered updates into reports, building onboarding checklists, reconciling spreadsheets. Anthropic built a coding agent and the office managers took it.

Claude Tag pushes this even further. In a video walkthrough, Anthropic showed Tag sessions that run for weeks — checking in daily, sending bug fixes, posting data readouts to Slack without anyone prompting it. The pitch is that you tell Claude something once and it remembers forever. Tell it to only monitor a specific type of issue in a channel and it adjusts. Tell it to answer every data question and it just does that going forward. Internally, Anthropic says Tag now accounts for 65% of the product team's pull requests.

Mobile is lighter than desktop by design — no local file access or browser control, just monitoring and approving what Claude already built. But the catch across all of this is access. Cowork on mobile requires a Max subscription starting at $100 a month, putting the full agent experience out of reach for most individual users. Claude Tag is in beta and Slack-only. And there's the trust question: Anthropic was recently caught fingerprinting Claude Code users without clear disclosure, and expanding into your phone and your company Slack raises fair questions about what data flows where.

If your team already pays for Claude Max and lives in Slack, Claude Tag is worth trying immediately. The Cowork mobile expansion is a nice convenience but not a reason to upgrade on its own. For most people at $20 a month on Pro, nothing here changes the value equation yet. But Anthropic is clearly building toward a version of Claude that doesn't wait for you to ask. When it gets cheaper and lands on more platforms, that's when it matters for everyone.

EVERYTHING ELSE IN AI

+ A 15-year-old in Japan used ChatGPT to build attack software that deleted 46,812 anime streaming accounts

+ China will let Alibaba, ByteDance, and DeepSeek buy Nvidia chips again but fewer than 200,000 total, and only with government approval

+ Meta's new AI tool lets anyone manipulate strangers' Instagram photos without notifying them, sparking immediate backlash

+ Researchers tricked GitHub's AI agent into leaking private code no coding skills, credentials, or access required

+ OpenAI's chief futurist is leaving after nine years , saying "the world is in on the secret" and he can now work on the mission from outside

+ Illinois just signed a frontier AI safety law requiring third-party audits, joining California and New York in building a de facto national framework without Washington

+ AI chip startup SambaNova raises $1B at an $11B valuation five months after its last mega round

+ The FTC proposed a new policy targeting companies that deliberately make their AI systems less accurate

OTHER TOOLS

+ Wispr Flow: *(sponsored)*: The new viral voice-to-text AI for iPhone and Mac that actually understands what you're saying every time

+ Superpower: *(sponsored)*: The health app that tests your blood and uses AI to tell you what's actually going on with your body before your doctor does

+ Slack: Slackbot can now pull your CRM data, generate charts, and send DocuSigns — all from a chat message, thanks to a new deep integration with Salesforce

+ Grok 4.5: SpaceXAI's new flagship model is live in Cursor and its API, with a 500K-token context window and built-in web search, code execution, and Office add-ins for Word, PowerPoint, and Excel

+ Meta Muse Image: Meta's new image generator can search the web for details you leave out of your prompt and write code to nail tricky visuals like charts — then revise its own output before you ever see it

+ Mistral Robostral Navigate: Mistral's new model lets robots navigate real-world spaces using just a single standard camera and plain-language instructions like "go to the second shelf on the left" — no expensive sensor arrays required

+ Voiskey: A voice typing app that turns your messy, rambling speech into clean, polished text by figuring out what you actually meant to say — not just transcribing what you said