ChatGPT vs Claude vs Gemini: Which One for Your Coursework?
Comparisons we found in search results rarely show dated prices or checked references. We collected the evidence that matters for papers: how often each chatbot invented references in a published same-prompt study, what each price step buys, what you pay where you live, and how to keep your chats out of training. Every claim about the chatbots has a source; the screenshots are our own, taken on Oct 12, 2026.
Janosch & Dr. Jonathan SchulteStudyTexter editorial team
Updated
21 min readDisclosure: This comparison is published by StudyTexter, a company in the same market. We only state verifiable facts with source and date. Method
Contents
There is no overall winner. Gemini is the cheapest of the three for U.S. college students this school year (Google AI Pro free for a year, redeem by Dec 31, 2026) and made up the fewest references when a study gave all three the same prompt (12%, against ChatGPT 38% and Claude 64%, partly read from the study’s chart; free versions, February 2025). Claude lets you keep your chats out of training with one setting and documents generous upload limits (20 files per chat, PDFs up to 1,000 pages). ChatGPT sits in the middle on this evidence; its student offer is shorter and only for U.S. schools (claim by Oct 31, 2026). The paid tiers cost about $20 a month in the US. None of the three gave fully reliable references, so for the paper itself use a tool that shows where each citation comes from, such as StudyTexter.
ChatGPT vs Claude vs Gemini: same prompt, references checked in a study
For a paper, the first question is simple: how many of the sources a chatbot names actually exist? We found one study that gave ChatGPT, Claude and Gemini the same prompt and checked every answer by hand: researchers at the Universidad Internacional de La Rioja tested the free versions of eight chatbots from February 7 to 9, 2025[1]. This was the prompt, word for word:
“I am a university student working on my Final Degree Project. I need you to provide me with 10 relevant academic references in the field of [your field]. Please format the references in APA 7th edition.”
Each chatbot answered it for five fields (cardiology, mechanical engineering, organic chemistry, sociology, art history). Here is how the three compare:
| Chatbot (free version, model) | Correct / partly correct / wrong or made up |
|---|---|
| Gemini Gemini Flash 2.0 | 32% / 56% / 12% |
| ChatGPT GPT-4o-mini | 30% / 32% / 38% |
| Claude Claude 3.5 Sonnet | 2% / 34% / 64% |
Across all 400 references from eight chatbots, only 26.5% were fully correct and 39.8% were wrong or made up[1]. 90% of Gemini’s references were books, and made-up journal articles were far more common than made-up books. Perplexity (72%) and Copilot (100%) did worse than all three; Grok and DeepSeek invented none[1].
Check your own chatbot with the same prompt
You do not have to trust a study from 2025. Copy the prompt, use your own field, check the references and enter what you found:
Fully correct in your test: . In the study: Gemini 32%, ChatGPT 30%, Claude 2%.
Our own test
- Prompt The study prompt, word for word, for psychology, business and nursing, in each free version, with web search off and on.pending
- Check Every reference via doi.org, Google Scholar and the publisher page.pending
- Score Fully correct, partly correct, not found, per chatbot.pending
- Repeat Three runs per chatbot on different days.pending
Pending: the test needs accounts and manual work, and we do not operate other companies’ tools with scripts. Until then this page relies on the published study.
Which chatbot for which coursework task
A chatbot that explains well can still fail at your bibliography. So here is a pick for each task students actually have, each with its evidence. Where the evidence is thin, we say so.
| Coursework task | Best documented fit and why |
|---|---|
| Understand a lecture or a hard reading | any of the three, free All three have a free plan[4][5][17]. Google’s US student page adds study notebooks and quizzes from your own material[13]. |
| Work through several long PDFs | Claude; Gemini with AI Pro Claude: up to 20 files per chat, PDFs up to 1,000 pages (text only above 100)[20]. Gemini AI Pro: 1M-token context[21]. ChatGPT: 512 MB per file; Free: 3 uploads a day[18]. |
| Get a reading list | none of them on its own Gemini made up the fewest references in the same-prompt study (12%), Claude the most (64%)[1]. Check every one. |
| Style of your own writing | Claude or Gemini, by a public vote Google and Anthropic models held the top 15 places of a creative-writing leaderboard on Oct 8, 2026[26]. OpenAI’s best model ranked 20th (statistical range 10 to 37). That is not academic feedback: test with your own paragraph. |
| Keep unpublished data private | any of the three, with training switched off All three let you opt out of training; Claude’s incognito chats and the temporary chats of ChatGPT and Gemini are not used for training[23][22][25]. Check the setting before you upload. |
| Lowest cost for a full school year (US) | Gemini Google AI Pro free for a year for U.S. college students (redeem by Dec 31, 2026)[13]; ChatGPT: 4 months of Plus (claim by Oct 31, 2026)[16]; Claude: no individual student price listed on its pricing page[4]. |
If you want a broader look at tools for academic writing, see ChatGPT alternatives for academic writing and StudyTexter vs other AI tools.
Free, $5 to $8, $20: what each price step buys
All three have a free plan, and all three charge about $20 a month for the plan most students consider. The differences are in the steps below and above that, and in the limits:
| Price step (US) | What each provider’s page says |
|---|---|
| $0 | ChatGPT Free: about 12 pages of input per message, 3 file uploads a day, limited deep research[17][18] Claude Free: web search, up to 5 projects, no Research; a usage cap per session that resets every five hours[4][19] Gemini Free: 32K-token context, Deep Research (may be unavailable at peak times), model 3.5 Flash-Lite[21][5] |
| $5 to $8 | ChatGPT Go: $8 at launch; about 40 pages of input, about 320 pages with reasoning; in January 2026 OpenAI said it planned to test ads in Free and Go in the US[7][17] Claude no plan between Free and Pro on the pricing page[4] Gemini Google AI Plus: $4.99; 128K context, 2x the free usage, access to 3.1 Pro[5][21] |
| about $20 | ChatGPT Plus: $20; expanded deep research, projects, about 320 pages with reasoning[6][17] Claude Pro: $20 monthly or $17 a month billed yearly ($200), plus tax; more usage, Research, projects (Free: up to 5)[4] Gemini Google AI Pro: $19.99; 1M context, 4x the free usage, Gemini in Docs, 5 TB storage[5][21] |
| $100 and up | ChatGPT Pro from $100[8]; Claude Max from $100[4]; Google AI Ultra from $99.99[5]. Rarely needed for coursework. |
- 1Free, $0: includes web search, files and code.
- 2Pro: $17 a month with the yearly plan ($200 up front), $20 if billed monthly.
- 3Max: from $100 a month.
- 4Fine print: prices shown don’t include tax; usage limits apply.
Screenshot of claude.com/pricing, Oct 12, 2026, 11:19 UTC, taken by StudyTexter without signing in; news pop-up hidden, cropped. © Anthropic.[4]
Claude has no cheap middle step: after Free comes Pro. Google’s ladder starts lowest, at $4.99 for AI Plus, and its $19.99 AI Pro comes with a 1-million-token context[5][21].
- 1Country view: United States; other countries show their own currency.
- 2Google AI Plus: $4.99 a month, 2x the free usage.
- 3Google AI Pro: $19.99 a month, 4x the free usage.
Screenshot of gemini.google/subscriptions, Oct 12, 2026, 11:19 UTC, taken by StudyTexter without signing in; cookie notice hidden (not accepted), cropped. © Google.[5]
What ChatGPT, Claude and Gemini cost where you study
Many readers of our English pages live outside the US, and the providers handle that very differently. Google shows a local price on each country page we opened. OpenAI bills in 68 currencies but shows the amounts only on its pricing page. Anthropic shows USD and GBP prices; in some regions it bills in local currency (in India, for example, according to a press report[11]), which we could not see from our US connection, so check the price at checkout:
| Where you pay | The about-$20 tier, per month as shown |
|---|---|
| United States | ChatGPT Plus $20[6] · Claude Pro $20 (or $17 billed yearly) plus tax[4] · Google AI Pro $19.99 (AI Plus $4.99)[5] |
| United Kingdom | Claude Pro £18 plus tax[10] · Google AI Pro £18.99 (AI Plus £4.49)[9] · ChatGPT bills in GBP, amount not visible to us[12] |
| India | Google AI Pro ₹1,950 (AI Plus ₹399)[9] · Claude Pro ₹2,000 including GST, according to a press report[11] · ChatGPT bills in INR, amount not visible to us[12] |
| Philippines | Google AI Pro ₱1,100 (AI Plus ₱285)[9] · ChatGPT bills in PHP[12] · Claude: local price not shown to us (US connection) |
| Italy, Germany (euro) | Google AI Pro €21.99 (AI Plus €4.99)[9] · ChatGPT bills in EUR[12] · Claude: euro price not shown to us (US connection) |
| Poland | Google AI Pro 97.99 zł (AI Plus 23.99 zł)[9] · ChatGPT bills in PLN[12] · Claude: złoty price not shown to us (US connection) |
| Australia, Canada | Google AI Pro A$32.99 or CA$26.99 (AI Plus A$7.99 or CA$6.99)[9] · ChatGPT bills in AUD and CAD[12] |
In India and the Philippines, Google AI Plus costs ₹399 or ₱285 a month[9], the lowest paid AI plan on Google’s page there. Do not convert prices yourself to compare them; tax and the price your bank charges can change the amount at checkout.
ChatGPT vs Gemini: the student offers
For students, the biggest difference between ChatGPT and Gemini this school year is the free time you get:
- Gemini (US): Google AI Pro free for one year for college students 18 and older; redeem by Dec 31, 2026; a card is needed, and after the year it renews at $19.99 a month unless you cancel[13]. Outside the US, Google’s student plan is usually AI Plus (in the UK, for example)[14], and in Canada the student offer ended on Dec 9, 2025[15]; see Gemini for students by country.
- ChatGPT (US only): four free monthly billing periods of Plus for students at eligible U.S. degree-granting institutions; claim by Oct 31, 2026; a payment method is needed, and it then renews at $20 a month unless you cancel[16].
- Claude: no student price for individuals on Anthropic’s pricing page[4].
- 1The plan: the free Google AI Pro student plan (one year, per the page headline).
- 2Deadline: redeem until December 31, 2026.
- 3Card needed: a valid form of payment at sign-up.
- 4After the year: $19.99 a month unless you cancel.
Screenshot of gemini.google/students, Oct 12, 2026, 11:20 UTC, taken by StudyTexter without signing in; cookie notice hidden (not accepted), cropped. © Google.[13]
Beyond the price, Gemini fits best if you already work in Google Docs and Drive: Google lists Gemini in Docs as part of AI Pro[5]. All offers and their renewal dates, by country, are in our overview AI student discounts by country.
Claude vs ChatGPT for writing
“Which writes better” is hard to measure, and we have not run our own writing test. The largest public signal is Arena, where users compare two anonymous answers to the same prompt and vote. In its creative-writing category (updated Oct 8, 2026), ranks 1 to 15 were all Google and Anthropic models; OpenAI’s best model, gpt-5.6-sol-xhigh, ranked 20th, with a statistical range of 10 to 37[26]. These are model versions from the providers’ APIs; the free apps may run other models.
- 1Category: Creative Writing in the Text Arena.
- 2Updated Oct 8, 2026, based on users’ votes.
- 3Ranks 1 and 2: a Google model (1519 ±20, marked “Preliminary”) and an Anthropic model (1517 ±17), within each other’s margin.
Screenshot of arena.ai, Oct 12, 2026, 11:16 UTC, taken by StudyTexter without signing in, cropped. Not shown: OpenAI’s best model at rank 20 (statistical range 10 to 37).[26]
Two limits: the top two are within each other’s margin of error, and a model listed as gemini-3.5-flash-lite (Google names 3.5 Flash-Lite for the free Gemini plan) ranked 62nd[26][5]. A leaderboard shows which model people preferred for stories and poems. Feedback on a lab report is a different task.
A fair test you can do in ten minutes: take one paragraph you wrote, paste it into each chatbot with the same request (“Point out unclear sentences and weak arguments. Do not rewrite the paragraph.”), and compare what each one finds. Keep your own text and decide each change yourself.
Privacy: who trains on your chats
For papers you often paste in material that is not public: interview notes, unpublished data, drafts. All three let you keep chats out of training, but you have to check the setting yourself.
| Chatbot | Training on your chats (personal accounts) |
|---|---|
| Claude | Training: only if you allow it in “Model Improvement”, says Anthropic’s privacy page (chats flagged for a safety review are an exception); its pricing page lists model training as “Opt-out” for all personal plans, so check what the setting shows[23][4] If allowed: chats kept up to 5 years, otherwise 30 days[24] Incognito chats: never used for training[23] |
| ChatGPT | Training: “we may use your content to train our models”, and you can opt out[22] Switch off: Settings › Data controls › “Improve the model for everyone”[22] Temporary chats: not used for training[22] |
| Gemini | With Keep Activity on: chats can be used to improve Google’s models, and some are reviewed by people, including Google’s service providers; reviewed chats are kept up to 3 years[25] Switch off: turn off Keep Activity; chats are then kept for 72 hours[25] Temporary chats: not used for training[25] |
If you use them: citing ChatGPT, Claude or Gemini in APA or MLA
Many instructors set their own AI rules in the syllabus, so check them first. If you may use a chatbot and quote or paraphrase its output, both major US styles have guidance:
- APA: the company is the author (OpenAI, Anthropic or Google). Since APA’s September 2025 update you can cite the specific chat with its date and shared link; keep a record of your prompts, and long answers can go in an appendix[27].
- MLA: do not treat the AI as an author. Describe what it generated, including the prompt, name the tool as the container, add the version, the company and the date you generated it[28].
One rule beats both styles: if a chatbot tells you a fact from a paper, find the paper and cite the paper, not the chatbot.
Where ChatGPT, Claude and Gemini are better than StudyTexter
All three do many things StudyTexter does not, and two of them are free to start. Honestly split:
Use ChatGPT, Claude or Gemini when you …
- want a concept, method or statistic explained
- turn your notes into flashcards or practice questions
- summarize long readings or several PDFs (Claude documents up to 20 files per chat)
- need code, tables or charts for a data analysis
- want quick feedback on a paragraph you wrote
Use StudyTexter when you …
- write a research paper or thesis with real sources
- want to accept or reject each source yourself, from over 479 million
- want every citation with its source and page number in the Evidence View
- need a citation style from over 2,800, chosen in step 3 (free)
- prefer paying once to a monthly subscription
For the paper itself: StudyTexter
A chatbot answers what you ask. A research paper needs a question, an outline, sources that exist and citations you can back up. StudyTexter guides you in 12 steps from your topic to a full draft, in up to 2 hours. In the source check you accept or reject each suggested source, and in the draft every citation shows its source and page number. You revise the draft and stay the author.
| Feature | Chatbots |  |
|---|---|---|
| Sources | from the model or web search; 12 to 64% wrong or made up in one same-prompt study (free versions, Feb 2025) | 479M+ sources; you accept or reject each one |
| Citations | page numbers, if given, need checking against the source | source and page number for each citation in the Evidence View |
| Workflow | open chat | 12 steps from topic to export |
| Citation styles | whatever you ask for in the prompt | over 2,800, chosen in step 3 (free) |
| Checks | plagiarism and AI check not mentioned on the pricing pages | plagiarism and AI check included |
| Export | copy the text or ask for a file | Word, PDF, LaTeX, Google Docs |
| Cost | free plans; about $20 a month for the paid tiers | steps 1 to 5 free, then $79 once for 20 pages |
“Losses loom larger than gains.”
Steps 1 to 5 are free, up to your proposal draft; the source check (step 6) and everything after it need a one-time package from $79.
ChatGPT vs Claude vs Gemini: FAQ
Is Claude better than ChatGPT?
It depends on the task. Claude lets you switch training off (setting “Model Improvement”), never trains on incognito chats and documents generous upload limits (20 files per chat, PDFs up to 1,000 pages), and Anthropic models ranked high on a public creative-writing leaderboard in October 2026. For reference lists, the free Claude did worse than the free ChatGPT in a February 2025 study with the same prompt: 64% of its references were wrong or made up, against 38% for ChatGPT (the ChatGPT value read from the study’s chart).
Is Claude better than ChatGPT for writing?
On the Arena creative-writing leaderboard updated Oct 8, 2026, ranks 1 to 15 were all Google and Anthropic models and OpenAI’s best model ranked 20th (statistical range 10 to 37), with overlapping margins. That is a crowd vote on API models, not a test of the apps or of academic writing. Paste the same paragraph of your own into both and judge the feedback yourself.
Gemini vs ChatGPT: which is better for students?
In the US, Gemini currently offers more free time: college students get Google AI Pro free for a year if they redeem by Dec 31, 2026, while OpenAI gives four free months of ChatGPT Plus to students at U.S. schools who claim by Oct 31, 2026. In the same-prompt study, the free Gemini also made up fewer references (12%) than the free ChatGPT (38%; both read from the study’s chart).
Is Claude free?
Yes: the free plan includes web search and up to 5 projects, with a usage cap that resets every five hours. Claude Pro costs $20 a month or $17 a month billed yearly in the US, plus tax; see the price table above.
Which AI makes up the fewest sources?
In the one same-prompt study we found (free versions, February 2025): Gemini 12%, ChatGPT 38%, Claude 64% wrong or made up (the study states Claude’s value; the other two we read from its chart). None was fully reliable, so check every reference.
How do I cite ChatGPT, Claude or Gemini in APA or MLA?
In APA, the company is the author (OpenAI, Anthropic or Google); describe the prompt, and since APA’s September 2025 update you can cite the specific chat with its date and link. In MLA, the AI is not the author: the prompt describes the source, the tool is the container, and you add the date. Ask your instructor which style and disclosure they expect.
Which AI should I use for my research paper?
Use the chatbots to understand material and get feedback. For the paper itself, StudyTexter takes you in 12 steps from topic to draft: you accept or reject each source from over 479 million, and every citation shows its source and page number in the Evidence View. Steps 1 to 5 are free; then a one-time payment from $79. You stay the author.
Method and disclosure
We read websites, terms and reviews ourselves, back every number with a source and access date, and only add test results once we have them. Found a mistake? We correct it with a date.
On Oct 12, 2026 we read the pricing, help, student and privacy pages of OpenAI, Anthropic and Google in their US versions, and Google’s subscription pages for the UK, India, the Philippines, Italy, Germany, Poland, Australia and Canada. Screenshots were taken between 11:16 and 11:20 UTC without signing in; originals are on file with checksums. chatgpt.com, help.openai.com and openai.com showed a security check to our browser; we stopped there and used the text of those pages. Our requests ran through a US data center, so pages showed US prices.
For the citation section we used the one published study that gave all three chatbots the same prompt (February 2025, free versions) and a 2026 preprint on cited links. The study has since appeared in a journal{q('unir_j')}; we used the figures of the preprint. Shares the study gives only as a chart we measured from the chart. We created no accounts and ran no test of our own.
StudyTexter sells a guided draft for research papers and has an interest in this topic. That is why every statement about the chatbots has a source and date and we say plainly where the chatbots are better. ChatGPT, Claude and Gemini are trademarks of their owners. If you find a mistake, we will correct it and date the change.
Changes: Oct 12, 2026 first version; the same day, revised after an independent fact check (privacy wording for all three, Claude local prices, chart values marked, APA update, Arena screenshot, shorter page).
Sources
- Assessing the performance of 8 AI chatbots in bibliographic reference retrieval, Universidad Internacional de La Rioja (preprint, arXiv, May 23, 2025)free versions, same prompt, 400 references, tests Feb 7 to 9, 2025; per-chatbot shares read from Figure 1 · accessed Oct 12, 2026
- The same study as a journal article: Journal of Data and Information Science 11(2), 2026, pp. 342 to 354 (UNIR repository)DOI 10.1515/JDIS-2025-0326 · accessed Oct 12, 2026
- Detecting and Correcting Reference Hallucinations in Commercial LLMs and Deep Research Agents, University of Pennsylvania (preprint, arXiv, Apr 3, 2026)3 to 13% of cited URLs never existed, 5 to 18% did not resolve · accessed Oct 12, 2026
- claude.com: Pricing (US view)Free $0; Pro $17/month billed yearly ($200) or $20 monthly; Max from $100; tax not included · accessed Oct 12, 2026
- gemini.google: Subscriptions, United StatesFree $0, Google AI Plus $4.99, Google AI Pro $19.99, Ultra from $99.99 per month · accessed Oct 12, 2026
- OpenAI Help Center: What is ChatGPT Plus?$20/month · accessed Oct 12, 2026
- OpenAI: Introducing ChatGPT Go (Jan 16, 2026)$8/month in the US at launch; localized in some markets · accessed Oct 12, 2026
- OpenAI Help Center: What is ChatGPT Pro?Pro at $100 (5x Plus) and $200 (20x Plus) · accessed Oct 12, 2026
- gemini.google: Subscriptions for the UK, India, the Philippines, Italy, Germany, Poland, Australia and CanadaAI Plus and AI Pro prices in local currency · accessed Oct 12, 2026
- Claude Help Center: How much does Claude Pro cost?$20 (US) or £18 (UK) plus applicable tax · accessed Oct 12, 2026
- Business Today: Anthropic introduces India pricing for Claude AI plans (Jul 14, 2026)press report: Pro Rs 2,000/month including GST · accessed Oct 12, 2026
- OpenAI Help Center: Multi-currency billing68 billing currencies; amounts only on the pricing page · accessed Oct 12, 2026
- gemini.google: Gemini for Students, United StatesGoogle AI Pro free for 1 year, redeem until Dec 31, 2026, then $19.99/month · accessed Oct 12, 2026
- gemini.google: Gemini for Students, United Kingdomstudent plan: Google AI Plus, then £4.49/month · accessed Oct 12, 2026
- gemini.google: Gemini for Students, Canada“Our previous student offer ended on December 9, 2025” · accessed Oct 12, 2026
- OpenAI Help Center: ChatGPT Back to School offer for students4 free months of Plus, U.S. schools only, claim by Oct 31, 2026 · accessed Oct 12, 2026
- chatgpt.com: Pricing (plan comparison; amounts load by script and were not visible to us)context window and input maximum per plan · accessed Oct 12, 2026
- OpenAI Help Center: File Uploads FAQ512 MB per file; Free: 3 uploads per day · accessed Oct 12, 2026
- Claude Help Center: About free Claude usagesession limit, resets every five hours · accessed Oct 12, 2026
- Claude Help Center: Upload limits (updated Jul 23, 2026)500 MB per file, 20 files per chat, PDFs up to 1,000 pages · accessed Oct 12, 2026
- Gemini Apps Help: Usage limits and upgradescontext 32K without a plan, 128K AI Plus, 1M AI Pro · accessed Oct 12, 2026
- OpenAI Help Center: How your data is used to improve model performanceconsumer content may be used for training; opt-out · accessed Oct 12, 2026
- Anthropic Privacy Center: Is my data used for model training? (updated Mar 16, 2026)only if you allow it; incognito chats never · accessed Oct 12, 2026
- Anthropic: Updates to Consumer Terms and Privacy Policy (Aug 28, 2025)5-year retention if you allow training, otherwise 30 days · accessed Oct 12, 2026
- Gemini Apps Privacy Hub (updated Sep 24, 2026)Keep Activity, human reviewers, up to 3 years, 72 hours when off · accessed Oct 12, 2026
- Arena (formerly LMArena): Creative Writing leaderboard (updated Oct 8, 2026)anonymous side-by-side votes on API models · accessed Oct 12, 2026
- APA Style blog: How to cite ChatGPT (archived; APA published updated guidance for generative AI on Sep 9, 2025)the company is the author; describe the prompt · accessed Oct 12, 2026
- MLA Style Center: How do I cite generative AI in MLA style? (revised August 2025)AI is not an author; prompt as title, tool as container · accessed Oct 12, 2026
Read next
Brands and logos mentioned belong to their owners and are used for identification only. They are not affiliated with StudyTexter.