Don’t Waste Money: Which AI Subscription Is Worth It?

Product(s) reviewed

The video compares four AI subscription tools available/used in 2026: ChatGPT, Claude, Gemini, and Perplexity, focusing on what each is best at, privacy concerns, free-tier limits, and whether upgrading is worth it.

---

ChatGPT (OpenAI)

What it’s good at

• Day-to-day AI tasks: brainstorming, document analysis, reasoning
• Improved image generation
• Image + file + Google Drive-style workflow: upload photos/files and use connectors/app integrations
• Features mentioned:
  • Deep research
  • Agent mode
  • Study & learn
  • Web search
  • Canvas
• Model noted: defaults to GPT 5.2
• Context window: ~400,000 tokens (larger documents and long work sessions)
• User experience:
  • “Thinking” mode (on Pro) tends to handle large amounts of info across longer chats better than prior models

What it’s not so great at (cons)

• Slow responses in “thinking mode” (reported ~45+ seconds for outputs that should take less)
• More rigid prompting requirements since GPT 5; outputs not as good as older models
• Limited compared to Gemini on context: Gemini has 1M tokens

Privacy concerns / data use

• By default, chats and uploads are used to train unless opting out via Data Controls (“improve the model for everyone” toggle off)
• If you connect third-party apps (e.g., Google Docs), their privacy rules apply
• Caution for privacy when using random MCP/app servers (third-party integrations)

Free plan limitations

• No advanced reasoning (like 5.2 thinking)
• Message limits: roughly 10 messages per ~3 hours
• Limited voice, capped image generation, limited deep research
• No access to custom GPTs

Upgrade verdict (ChatGPT)

• Worth upgrading if you want custom GPTs
• Otherwise, the reviewer says there are better AIs worth the money

Unique points mentioned

• GPT 5.2: 400k token context
• Access to “modes” including auto vs thinking, plus legacy models
• Ability to create/connect apps via MCP/app integrations and developer mode

---

Claude (Anthropic)

What it’s good at

• Content creation (“hands down” best per reviewer)
• Nuanced conversation, ethical reasoning, analysis
• Coding: Claude Code praised as “amazing”
• Strong business/workflow positioning (“business go-to”)
• Rich integrations and workflow tools:
  • Connectors/MCP integrations are described as robust
  • Options include connecting to tools like Asana, HubSpot, etc. (and more listed for desktop extensions)
  • Desktop app emphasized as a major differentiator
• Artifacts: create apps quickly without coding
• Claude skills:
  • Turn on prebuilt skills or use skill creator to generate new skills via chat

What it’s not so great at (cons)

• Smaller context window: about 200,000 tokens
  • Exception: API/beta users can reach up to 1 million tokens
• Limited multimodal: can’t generate images (at least as stated)

Privacy / data use

• Anthropic positioned as more responsible; default training help is on
• To opt out: privacy setting toggle (“help improve Claude”)

Free plan limitations

• Daily usage limits
• No access to Opus 4.5 or extended thinking
• No Claude Projects
• Very limited connectors

Upgrade verdict (Claude)

• Reviewer strongly recommends upgrading for:
  • Content creators
  • Coders
  • Users solving complex problems
• Otherwise, free may be fine for light users

Unique points mentioned

• Desktop app workflow is a “game changer” for non-technical file access
• Artifacts for rapid MVP-style builds
• Skills/skill creator as a key differentiator

---

Gemini (Google)

What it’s good at

• Gemini 3 Pro described as excellent; reviewer now uses it almost daily
• Massive context window: 1 million tokens (best for large documents/long conversations)
• Strong multimodal: image generation, video generation (Veo-3), audio mentioned
• Strong for coding, brainstorming, problem-solving
• Deep research with sources; integration with Google ecosystem:
  • Can search/upload across Gmail/Drive/Chat if connected
• “Gems” (similar to GPTs/projects): create preconfigured assistants with default tools (e.g., deep research by default)
• Tool features noted:
  • NotebookLM (upgraded on paid)
  • Canvas for creating presentation slides

What it’s not so great at (cons)

• Content creation not as strong as Claude (responses feel “dry”/less personality)
• Lack of MCP connections is frustrating
  • Few app connections listed (e.g., GitHub, Google Workspace tools, YouTube Music, and a handful like Asana/HubSpot/Mailchimp)

Privacy

• Follows Google privacy policy
• Chats/uploads/voice/screen recordings used for training unless opt out via Gemini Apps Activity

Free plan limitations

• Lower usage caps across chats, prompts, deep research, uploads, image generation
• Free context window reported as ~32,000 tokens vs 1M on paid

Upgrade verdict (Gemini)

• Previously “no”; now yes, because Gemini 3 Pro is “that good”

Unique points mentioned

• Video/image generation tools: Nano Banana Pro (images) and Veo-3 (video)
• Canvas highlighted as underused
• Biggest technical edge: 1M token context

---

Perplexity

What it’s good at

• Fast, real-time search with citations
• Strong for quick questions where you want sources linked per bullet
• Source controls:
  • web, social discussions, SEC filings, academic research
  • can include business tools like Slack/Notion/Gmail/Google Drive
• Paid feature callouts:
  • Tasks: schedule recurring searches and receive results
  • Spaces: organize research into projects (similar to ChatGPT projects)
  • MCP/connectors exist but are described as more limited
• Connector advantage stated:
  • For major connectors (e.g., Google tools), has read and write capabilities

What it’s not so great at (cons)

• Deep research depth not as strong as Gemini’s deep research (and compared to “Gen Spark” in the narration)
• Reviewer is “seldom impressed” with answers overall due to quality limitations

Privacy / data use

• Queries/chats logged and used to train by default unless opting out

Free plan limitations

• “Practically unlimited” basic searches
• Very limited pro searches (wording: ambiguous limitation)
• No access to advanced AI models or image generations

Paid pricing & tiers

• Pro: $20/month or $200/year
• Also mentions Perplexity Max for power users
• Mentions “Perplexity Assistant” in Max tier (email assistant, smart inbox labeling, drafting, calendar, etc.)

Upgrade verdict (Perplexity)

• Worth paying for if you need:
  • up-to-date info with citations
  • lots of academic research
• Otherwise, the reviewer does not recommend upgrading

Unique points mentioned

• Tasks + Spaces as standout workflow/research organization features
• Image generation supported on the platform (model choice: e.g., Nano Banana, Flux)

---

Overall comparisons / key takeaways

• Claude: best for versatile day-to-day use (reviewer’s final “pay for one” choice) and strongest emphasis on content creation + coding.
• Gemini: strongest technical advantage is 1M-token context and multimodal (images/video/audio), plus upgraded Google-integrated tools.
• ChatGPT: best value mainly if you specifically want custom GPTs; otherwise criticized for slower thinking mode and more rigid prompting.
• Perplexity: best for search with citations and recurring research, but weaker deep research depth and answer quality impressed less overall.

---

Main pros/cons highlighted per tool (condensed)

• ChatGPT
  • Pros: strong general reasoning; image generation improved; good integrations; huge (400k) context; deep research/agents
  • Cons: slow “thinking”; prompt rigidity; weaker output quality vs earlier models
• Claude
  • Pros: best content creation; strong coding; artifacts; skills; robust desktop/file workflow; connectors described as strong
  • Cons: smaller context (200k); limited multimodal (no image generation stated)
• Gemini
  • Pros: 1M tokens; very strong multimodal + deep research; coding/brainstorming; Google Workspace advantage; gems/canvas
  • Cons: weaker content creation personality; limited MCP connections
• Perplexity
  • Pros: fast web search with citations; tasks scheduling; spaces; connector read/write; strong for academic/web queries
  • Cons: deep research depth perceived as weaker; mixed overall answer satisfaction

---

Final verdict / recommendation

• If paying for just one tool: the reviewer recommends Claude as the most consistently useful.
• Choose based on needs:
  • Claude for content + coding + workflow versatility
  • Gemini for massive context and multimodal (incl. video) work
  • ChatGPT mainly for custom GPTs
  • Perplexity for fast, citation-heavy research and scheduled searches

---

Speakers/views

Only one main narrator/reviewer is present in the subtitles, but they discuss multiple aspects (model performance, UX features, privacy, free-tier vs paid) across the same speaker throughout.