Claude vs ChatGPT vs Gemini
Wins 3 of 9 rounds
Wins 1 of 9 rounds
Wins 4 of 9 rounds
There is no single winner; each is best at something different. Claude has the strongest models (Opus 5.5 scores 58 on the Artificial Analysis Intelligence Index) and wins for writing and coding. ChatGPT has the widest set of tools in one app and 900 million weekly users. Gemini is the best value, with a $4.99 plan, image and video creation and deep Google integration. All three main plans cost about $20 a month.
At a glance
| Claude | ChatGPT | Gemini | |
|---|---|---|---|
| Maker | Anthropic | OpenAI | |
| Top model in the app | Fable 5.1, Opus 5.5 | GPT-6 Astra (Pro plan) | Gemini 3.8 Flash; Deep Think on Ultra |
| AA Intelligence Index (best model) | 58 (Opus 5.5) | 53 (GPT-6 Astra) | 41 (Gemini 3.8 Flash) |
| Free plan | Sonnet and Haiku; no ads | GPT-5.6 Luna unlimited text; ads in US | 3.6 Flash, limited 3.1 Pro; no ads |
| Cheapest paid plan | Pro $20 ($17 yearly) | Go $8 | AI Plus $4.99 |
| Standard plan | Pro $20 | Plus $20 | AI Pro $19.99 |
| Heavy-use plans | Max $100 / $200 | Pro from $100 | AI Ultra $99.99 / $199.99 |
| Image / video creation | No / No | Yes / Not a core plan feature | Yes / Yes (from AI Plus) |
| Coding agent | Claude Code (Pro+) | Codex (all plans) | Google Antigravity |
| Users | Not disclosed | 900M weekly (Feb 2026) | 1B+ monthly (Aug 2026) |
Head to head
Model intelligence Winner: Claude
On the independent Artificial Analysis Intelligence Index, Claude Opus 5.5 scores 58, first of all models. OpenAI's GPT-6 Astra and Claude Fable 5.1 tie on 53. Google's best listed model, Gemini 3.8 Flash, scores 41. Google's promised flagship, Gemini 3.5 Pro, had still not launched by 23 September 2026, four months after it was announced.
Coding Winner: Claude
On Terminal-Bench 4.0 (real coding tasks in a command line), Anthropic reports 66.4% for Opus 5.5 and OpenAI reports 57.9% for GPT-6 Astra. Google has not published a Terminal-Bench 4.0 score for its app models. All three offer coding agents: Claude Code (from $20 Pro), Codex (every ChatGPT plan, so it is the cheapest to try) and Google Antigravity.
Writing and blind chat votes Winner: Claude
On LMArena's text leaderboard (13 September 2026), Claude Fable 5 ranked #1 (1506), Gemini 3.8 Flash #9 (1493) and the best OpenAI model listed, GPT-5.6 Sol, #18 (1483). GPT-6 had not yet been ranked. For long-form writing and a consistent voice, Claude remains the favourite.
All-in-one features Winner: ChatGPT
ChatGPT covers the most ground in one app: search, deep research, images, voice with live video, memory, projects, scheduled tasks, agent mode, Codex, ChatGPT Work for documents and slides, and the Atlas browser. Claude is strong on agents and Office files but has no image generation. Gemini matches much of it but keeps its best agent (Spark) and Deep Think for Ultra.
Images and video Winner: Gemini
Gemini makes images (Nano Banana 2) on every plan and videos (Gemini Omni) from AI Plus at $4.99. ChatGPT makes images on every plan, limited on Free, but video is not part of its standard plan list. Claude cannot generate images or video.
Free plan Winner: Gemini
Gemini Free includes image generation, Deep Research, Gemini Live and limited Gemini 3.1 Pro, with no ads, though its context window is only 32K tokens. ChatGPT Free has unlimited text chats with GPT-5.6 Luna but shows ads to US users. Claude Free is ad-free but uses smaller models and lower limits.
Price and value Winner: Gemini
Gemini starts at $4.99 (AI Plus, with 400 GB of storage) and AI Pro at $19.99 adds a 1M-token context window and more storage. ChatGPT's cheapest plan is $8 Go; Claude's is $20 Pro ($17 billed yearly). At the top end all three sit near $100 and $200. For most people who just want a capable assistant, Gemini costs least.
Integration with your phone and apps Winner: Gemini
Gemini is built into Android, Chrome, Gmail, Docs, Sheets and Google Search's AI Mode. ChatGPT and Claude connect to Google and Microsoft apps through connectors, and Claude has Microsoft 365 add-ins, but both are apps you open separately.
Ads and privacy Tie
Claude and Gemini show no ads in their apps. ChatGPT has shown ads to free US users since 17 January 2026 and says Go may show them. All three let you limit how your chats are used for training, and all three offer business plans that exclude company data. Claude and Gemini share this round; ChatGPT's free tier is the weak spot.
Which should you choose?
Choose Claude if…
- You write a lot and want the best-rated prose
- You are a developer and will use Claude Code daily
- You want the strongest model available on a $20 plan
- You want agents that hand back finished Word, Excel and PowerPoint files
Choose ChatGPT if…
- You want one app for research, images, voice, coding and documents
- You want unlimited free text chat and can live with ads
- You want to try a coding agent without paying
- You want ChatGPT Work to build slides, documents and spreadsheets
Choose Gemini if…
- You use Android, Gmail and Google Docs
- You want images and video on a cheap plan
- You want the best free plan with no ads
- You want cloud storage bundled with your AI
Claude and ChatGPT tie at the top for different reasons, with Gemini close behind as the value pick.
- Claude wins the quality rounds: best independent intelligence score, top of LMArena's blind votes and the lead on agentic coding. Pick it for writing, code and analysis. Skip it if you need images or a strong free plan.
- ChatGPT wins on breadth. It is the best single app if you do many different things and do not want to think about which tool to use. Skip it if ads on the free plan bother you.
- Gemini wins on price, media and integration. It is the best choice for Android and Google Workspace users and anyone on a budget. Skip it for hard coding and reasoning until Gemini 3.5 Pro ships.
If you can pay for one plan, pick by your main task. If you can pay for two, Claude Pro plus Gemini AI Plus ($24.99 total) covers writing, code, images and video.
Plans and prices side by side
| Tier | Claude | ChatGPT | Gemini |
|---|---|---|---|
| Free | $0, no ads | $0, ads in US | $0, no ads |
| Budget | None | Go $8 | AI Plus $4.99 |
| Standard | Pro $20 ($17 yearly) | Plus $20 | AI Pro $19.99 |
| Heavy use | Max $100 (5x) / $200 (20x) | Pro from $100 | AI Ultra $99.99 (5x) / $199.99 (20x) |
| Teams | Team $25/seat | Business (sales) | Google Workspace |
US web prices as of 23 September 2026. In-app prices on iPhone can be higher.
Benchmarks at a glance
| Test | Claude | ChatGPT (OpenAI) | Gemini (Google) | Source |
|---|---|---|---|---|
| AA Intelligence Index | Opus 5.5: 58 | GPT-6 Astra: 53 | 3.8 Flash: 41 | Artificial Analysis |
| LMArena text rank (13 Sep) | Fable 5: #1 | GPT-5.6 Sol: #18 | 3.8 Flash: #9 | LMArena |
| Terminal-Bench 4.0 | Opus 5.5: 66.4% | GPT-6 Astra: 57.9% | Not published | Anthropic; OpenAI |
| Flagship API price (per 1M tokens) | Opus 5.5: $4 / $20 | GPT-6 Astra: $10 / $50 | 3.1 Pro: $2 / $12 | Vendors |
Vendor benchmark figures are claims made at launch. Independent rankings change often as new models arrive.
What changed in 2026
- OpenAI moved ChatGPT through GPT-5.5, GPT-5.6 and, in September, GPT-6 Astra, Sol and Luna. Free US users started seeing ads in January.
- Anthropic released Claude Fable 5 in June, Fable 5.1 on 1 September and Opus 5.5 on 22 September, which beat Fable 5.1 at 40% of the price.
- Google launched Gemini 3.5 Flash at I/O in May, followed by 3.6, 3.7 and 3.8 Flash, cut AI Plus to $4.99 and added a $99.99 Ultra tier. The Gemini app passed 1 billion monthly users in August, but Gemini 3.5 Pro is still delayed.
- Pick by your main task, not by hype: writing and code point to Claude, many small mixed tasks to ChatGPT, Google apps and media to Gemini.
- Try all three free plans on the same three real tasks from your week before paying; the differences show quickly.
- On Gemini Free, split long files into parts because of the 32K-token context window; AI Pro raises it to 1M.
- On ChatGPT Free, turn off model training in Settings > Data Controls if you care about privacy.
- Claude's usage resets every five hours; plan long work sessions around that, or keep files in a Project so you do not re-upload them.
Jargon explained
- Artificial Analysis Intelligence Index
- An independent score that combines ten reasoning, knowledge, maths and coding tests into one number for each AI model.
- LMArena
- A public site where people vote blind between two AI answers; the votes become a ranking.
- Context window
- How much text an AI can keep in view at once, measured in tokens. One token is about three-quarters of a word.
- Coding agent
- An AI tool that can read code, edit files and run commands to finish a programming task on its own.
- Terminal-Bench
- A test where an AI has to complete real software tasks by typing commands into a terminal.
Frequently asked questions
Which is the best AI: Claude, ChatGPT or Gemini?
Claude has the strongest models on independent tests, ChatGPT has the most features and Gemini is the best value. See our best AI chatbots ranking for the full list.
Which is best for coding?
Claude. Opus 5.5 scores 66.4% on Terminal-Bench 4.0 against 57.9% for GPT-6 Astra, and Claude Code comes with the $20 Pro plan.
Which has the best free plan?
Gemini, for most people: image generation, Deep Research and Gemini Live with no ads. ChatGPT Free has unlimited text chat but shows ads in the US.
Which can make images and videos?
Gemini makes both. ChatGPT makes images. Claude makes neither, though it can draw charts and diagrams.
Are all three $20 a month?
The main plans are: Claude Pro $20, ChatGPT Plus $20 and Google AI Pro $19.99. ChatGPT also has an $8 Go plan and Google a $4.99 AI Plus plan.
Is Gemini 3.5 Pro available?
No. Google announced it in May 2026 but had not released it as of 23 September 2026. The newest Gemini Pro model is 3.1 Pro.
Sources
Every fact on this page comes from public information. Vendor figures are labelled as vendor claims.
- Claude pricing (Anthropic)
- ChatGPT pricing (OpenAI)
- Google AI Pro and Ultra plans (Google)
- Introducing Claude Opus 5.5 (Anthropic)
- GPT-6 Astra (OpenAI)
- Introducing GPT-6 Sol and Luna (OpenAI)
- Introducing Gemini 3.8 Flash and 3.8 Flash Cyber (Google)
- Gemini Apps limits and upgrades (Google)
- Gemini API pricing (Google)
- Google's Gemini app surges to 1 billion users (TechCrunch)
- Google releases three new Gemini models, but no 3.5 Pro (TechCrunch)
- LLM leaderboard: Intelligence Index (Artificial Analysis)
- Text Arena leaderboard (LMArena)
- ChatGPT (Wikipedia)