Updated 30 October 2026: OpenAI cancelled the lower Pro 200 allowance for new subscribers on 29 October, and Grok is now sold through The X Subscription, from $8 a month before VAT, so we have updated the ChatGPT Pro and Grok details.
Corrected 1 October 2026: Claude Pro reaches Fable 5.1 only through paid usage credits; Max includes it at half its weekly limit, according to Anthropic’s pricing page.
Which AI chatbot is best depends on the job. The one we’d hand a 300-page report isn’t the one we’d use to check sources, and neither is the obvious choice if you spend your day in Gmail. So this guide looks at eight of them, ChatGPT included, and tells you which to open for which job. It isn’t a rerun of our ChatGPT alternatives list, because here ChatGPT has to earn its place like everyone else. We read every vendor’s plan, model and help pages on 30 September 2026, the day after OpenAI’s DevDay.
- Best overall: ChatGPT. Images, voice, deep research and Codex in one app, and Plus is $20 a month.
- Best free: ChatGPT. Unlimited text chats with GPT-5.6 Luna. Gemini’s free plan has more extras, but it now runs only the lightweight Flash-Lite.
- Best for long documents: Claude. A 1M-token context window in chat on paid plans, and PDFs of up to 1,000 pages.
- Best for research with sources: Perplexity. Every answer shows its citations, though the free plan is tight.
- Best inside Google or Microsoft apps: Gemini or Copilot. Go with the office suite you already use.
- Best for privacy: Kagi Assistant. Kagi never trains on your chats, and plans start at $5.
The eight at a glance
Prices are US list prices for one person paying monthly. “Newest model in the app” is the newest one you can use in the vendor’s own consumer app, which isn’t always the newest it sells to developers.
| Chatbot | Newest model in the app | Free plan | Paid from | Pick it for |
|---|---|---|---|---|
 ChatGPTOpenAI · chatgpt.com | GPT-6.1 Sol, in Work and Codex only | GPT-5.6 Luna, unlimited text chats | $8 (Go), $20 (Plus) | Best overall Best free |
 ClaudeAnthropic · claude.ai | Sonnet 5.5 (28 Sep); Opus 5.5 on Pro | Sonnet and Haiku, web search | $20 (Pro), $17 on annual | Long documents |
 GeminiGoogle · gemini.google | Gemini 4 Argon on AI Pro and Ultra | Flash-Lite | $4.99 (AI Plus), $19.99 (AI Pro) | Most free extras Google apps |
 Microsoft CopilotMicrosoft · copilot.microsoft.com | Smart mode, “powered by GPT-5” | Chat, Smart mode, Think Deeper | $19.99 (Microsoft 365 Premium) | Office apps |
 PerplexityPerplexity AI · perplexity.ai | Sonar, plus third-party models on Pro | Basic search; 3 Pro Searches a day | $20 (Pro), $200 (Max) | Research |
 GrokSpaceXAI · grok.com | Grok 4.6 | Limited chat and images, live web and X search | $8 (X Subscription Plus) | Live news |
 DeepSeekDeepSeek · deepseek.com | DeepSeek-V4.1-Flash (10 Sep) | Web and app chat, no paid plan | No paid plan (API only) | Free model |
 Kagi AssistantKagi · kagi.com | GPT-6.1 Sol, Opus 5.5 and others on Ultimate | Trial of 100 searches | $5 (Starter) | Privacy |
How we chose
We didn’t run a benchmark or a blind test. We read each vendor’s plan page, release notes and help centre and compared the free plan, the first paid step, the model the app really runs, how much text one conversation holds, and whether your chats are used to train its models by default.
Two from our longer list didn’t make the cut. Meta AI lives inside WhatsApp and Instagram but has no paid plan and no training switch. Mistral Vibe is European and has a $14.99 Pro plan, but it doesn’t lead on any of the six jobs here.
The all-rounders: ChatGPT and Claude
ChatGPT
ChatGPT covers the most ground in one app. It writes, makes and edits images, runs deep research, talks back in Live voice and includes Codex for code. The free plan offers “unlimited text chats with GPT-5.6 Luna”, with a small 27K-token context window. Go costs $8 and, OpenAI says, may show ads. Plus is $20 on the web ($19.99 in the App Store) and is the plan we’d buy.
On 29 September OpenAI released GPT-6.1 Sol, which it says “nearly matches GPT-6 Astra’s intelligence” on coding, computer use and professional work at a fifth of Astra’s token prices. It’s rolling out to Plus, Pro, Business, Enterprise and Edu in ChatGPT Work and Codex, and OpenAI says it “is not yet available in Chat”. Pro now comes at $100, $200 (open to new subscribers again; on 29 October OpenAI cancelled the smaller allowance it had planned for them) and a new $500 tier that includes the faster Ultrafast mode.
Claude
Claude is the one we’d hand a long document. Anthropic’s help centre says Sonnet 5.5, Opus 5.5 and Fable 5.1 get a 1M-token context window in chat on paid plans, and you can upload 20 files per chat at up to 500 MB each, with PDFs up to 1,000 pages. That’s enough for a whole contract bundle in one conversation, whereas ChatGPT Plus stops at 54K tokens for Instant replies and 256K for reasoning.
The free plan runs Sonnet and Haiku with web search, file creation, code and memory. Sonnet 5.5 arrived on 28 September, and Anthropic says it’s “30%+ faster” than Sonnet 5 and “costs up to 30% less for most work”. Pro is $20, or $17 a month billed annually, and adds Opus 5.5 and Claude Code; Fable 5.1 costs extra usage credits on Pro, while Max includes it at half its weekly limit. Claude doesn’t generate images. If it’s down to these two, our Claude vs ChatGPT comparison goes deeper on memory, files and voice.
+Reasons to pick ChatGPT
- Images, Live voice and deep research in one app
- An $8 Go plan between Free and Plus
- Codex on Plus, where GPT-6.1 Sol lands first
- Unlimited text chats on the free plan
+Reasons to pick Claude
- 1M-token context in chat on paid plans
- PDFs up to 1,000 pages, 20 files per chat
- New models reach the chat menu on launch day
- You decide about training when you sign up
Built into Google and Microsoft
Gemini
Gemini’s free plan has the longest feature list here, though since 9 October free accounts get only Flash-Lite, Google’s lightest model. Google lists the Flash-Lite model, image generation, Deep Research, Gemini Live, Canvas and Gems, with a 32K-token context window. Google AI Plus at $4.99 moves you up to the Flash model, doubles your usage, raises context to 128K and puts Gemini in Gmail and Vids. AI Pro at $19.99 gives four times the free usage, a 1M-token window, Gemini in Docs and 5 TB of storage. AI Pro and Ultra also get Gemini 4 Argon alongside the Pro and Flash models.
If your mail and documents already live with Google, this is the easy choice. We set its $19.99 plan against the other two big ones in Claude, ChatGPT or Gemini.
Microsoft Copilot
Copilot is free on the web, in Windows and on mobile, and Microsoft says its Smart mode is “powered by GPT-5”. The real reason to pick it is the paid plan. Microsoft 365 Premium, at $19.99 a month or $199.99 a year, includes the Office apps and lets Copilot “create, edit, summarize, and analyze data across Word, Excel, PowerPoint” and Outlook, plus a Researcher agent for source-cited reports. If you already pay for Microsoft 365, start here.
For research and news
Perplexity
Perplexity is a search engine that answers in sentences and lists its sources under each reply, which is why it’s our research pick. The free plan is the weak spot: it offers “practically unlimited basic searches” but only three Pro Searches a day and one Research query a month. Pro is $20 in the App Store and adds third-party models alongside Perplexity’s own Sonar. Max is $200.
Grok
Grok searches the web and X in real time, so ask it about something that happened an hour ago. The free plan covers limited chat and image generation with web and X search. Since 12 October new subscribers buy The X Subscription, one plan for Grok and X: Plus at $8 a month adds Voice, Imagine and Build, Premium at $30 adds video generation and Grok Bot, and Super at $100 and Ultra at $200 raise the limits. Prices exclude VAT, and SuperGrok is now sold only to existing subscribers. SpaceXAI launched Grok 4.7 on 21 September for the API and Grok Build, but its consumer pricing page still lists Grok 4.6. We cover the rest in Grok vs ChatGPT.
Free, cheap or private
DeepSeek
DeepSeek gives you a very capable model for nothing, with no paid consumer plan at all. Since 10 September its chat has run DeepSeek-V4.1-Flash, a 552-billion-parameter mixture-of-experts model that DeepSeek says beats its own V4-Pro. The trade-off is your data: DeepSeek’s privacy policy says it’s stored in the People’s Republic of China, and the only way to opt out of training is by email. We explain what that means in practice in DeepSeek vs ChatGPT.
Kagi Assistant
Kagi Assistant comes with Kagi’s paid search and is our privacy pick. Kagi says “your data is never used to train AI models (not by us or by the LLM providers)”, and threads are deleted after 24 hours of inactivity by default. Starter is $5 a month with 300 searches and the Assistant in Quick mode, and Professional at $10 makes search unlimited. Ultimate at $25 adds Research mode and flagship models. Kagi’s docs list GPT-6.1 Sol, Claude Opus 5.5, Sonnet 5.5, Gemini 3.8 Flash and Grok 4.7 there, so you can chat with GPT-6.1 Sol on Kagi before ChatGPT’s own Chat view has it.
Verdict
If you pay for one, make it ChatGPT Plus at $20. It does the widest range of things, and since DevDay it gets GPT-6.1 Sol in Work and Codex, though ordinary chats don’t have it yet. Choose Claude Pro instead if your days are long documents or code, because the 1M-token context in chat makes a real difference. Gemini’s free tier is worth having on any phone, and AI Pro is the paid plan to get if you live in Gmail and Docs.
For narrower jobs, use Perplexity when you need sources, Copilot if you already pay for Microsoft 365 and Grok for breaking news. Pay Kagi $5 if you’d rather nobody trained on your chats. DeepSeek is a remarkable free model for anything you’d be happy to have stored in China.
Frequently asked questions
What is the best AI chatbot right now?
For most people it’s ChatGPT, because one $20 Plus plan covers writing, images, voice, research and coding. Claude is better for long documents, Perplexity for sourced research and Gemini if you live in Google’s apps.
What is the best free AI chatbot?
ChatGPT. The free plan gives unlimited text chats with GPT-5.6 Luna, a current OpenAI model. Gemini’s free plan has more extras, including Deep Research, Gemini Live, Canvas and image generation, but since 9 October it runs only the lightweight Flash-Lite model. Claude Free runs Sonnet and Haiku and is strong for writing, and DeepSeek’s free app runs V4.1-Flash but stores data in China.
Is GPT-6.1 Sol available in ChatGPT?
Only in parts of it. Since 29 September 2026 it has been rolling out to Plus, Pro, Business, Enterprise and Edu users in ChatGPT Work and Codex, starting with Pro. OpenAI says it is not yet available in Chat, and it isn’t listed for Free or Go.
Which chatbot can handle the longest documents?
Claude and Gemini, both with a 1M-token context window on paid plans. Claude takes PDFs up to 1,000 pages and 20 files per chat. Gemini needs AI Pro at $19.99 for the full window, and ChatGPT Plus tops out at 256K tokens for reasoning.
How much does a paid AI chatbot cost?
Usually about $20 a month. Claude Pro, ChatGPT Plus and Perplexity Pro are $20, and Google AI Pro and Microsoft 365 Premium are $19.99. Cheaper options are Google AI Plus at $4.99, Kagi Starter at $5 and ChatGPT Go at $8.
Sources and method
We checked prices, models and limits on the vendors’ own pages on 30 September 2026. We re-checked the Gemini models per plan on 28 October 2026. OpenAI’s pricing and help pages refused automated access that day, so ChatGPT’s 29 September changes come from OpenAI’s DevDay posts and help articles as read on 30 September, prices were confirmed against the US App Store listing, and the Free, Go and Plus plan details are from our 28 September check of OpenAI’s pricing page. Perplexity’s site also refused access, so its prices are from the App Store and its free limits from our 28 September reading of its help centre. We haven’t run a hands-on test of the eight for this guide, and the use-case picks are our reading of what each vendor documents.
- OpenAI, DevDay 2026 recap: Pro 500, dots, 29 September changes
- OpenAI, Introducing GPT-6.1 Sol: availability in Work and Codex, not yet in Chat, OpenAI’s comparison with GPT-6 Astra
- OpenAI Help Center, About ChatGPT Pro tiers: Pro 100, Pro 200 and Pro 500 prices and Ultrafast
- OpenAI Help Center, ChatGPT release notes: GPT-6.1 Sol rollout order, Pro 200 allowance cut cancelled on 29 October
- OpenAI, Introducing dots: plans and the EEA, Switzerland and UK exclusion
- OpenAI, ChatGPT pricing (checked 28 September): Free and Go features, context windows, training opt-out
- Apple App Store, ChatGPT: in-app prices for Go, Plus and Pro 100, 200 and 500
- Claude, Pricing: Free, Pro and Max prices and models
- Anthropic, Introducing Claude Sonnet 5.5 and Models overview: release date, speed and cost claims, current models, figure
- Claude Help Center, context window on paid plans and upload limits: 1M context by model, files per chat, PDF pages; Can Claude produce images? (read 28 September)
- Google, Gemini subscriptions (US) and Gemini Apps Help, limits and upgrades: plan prices, free-tier model, models per plan, usage multiples, context per plan
- Google, Introducing Gemini 3.8 Flash: consumer availability, figure
- Microsoft, Microsoft 365 Premium and Conversation modes in Microsoft Copilot: price, Copilot in Office apps, Researcher, GPT-5 Smart mode
- Apple App Store, Perplexity and Perplexity, Which subscription plan is right for you? (read 28 September): Pro and Max prices, free limits
- SpaceXAI, Grok pricing and News: X Subscription plans and prices, Grok 4.6 on consumer plans, Grok 4.7 date
- DeepSeek, V4.1-Flash release note, Models and pricing and Privacy Policy: model, date, API prices, data storage
- Kagi, Pricing and Kagi Assistant documentation: plan prices, model list, no-training statement, 24-hour deletion
- Mistral AI, Pricing: Vibe Pro price
- Training defaults (read 28 September): OpenAI, Data controls, Anthropic, Is my data used for model training?, Google, Gemini Apps Privacy Hub, Microsoft, Privacy FAQ for Copilot, Perplexity, Data Collection, SpaceXAI, Privacy Policy and Meta, generative AI privacy
