Claude vs. ChatGPT as a Language Tutor: 11 Prompts, One Surprising Winner

A free tutor still costs you money and time. Expect to pay $0 in fees and about 45 minutes of setup, review and retries across the eleven prompts below, because the checking is yours.

The test case is an intermediate Spanish learner facing a 10-minute work-and-travel talk. Fresh chats each time, both free tiers, nothing spent.

The winner is not the one you would guess. Claude handles the jobs that need a lot of text at once, heavy correction and uploaded documents. ChatGPT Free, or Go at $8 a month, wins fast daily chat and quick voice drills.

The $0 contest is not about which model is smarter. It is about how much text each one holds and what a free session gives you. Claude Free is ad-free. ChatGPT Free shows ads in Germany as of October 2026.

You get eleven prompts to paste as-is, a verdict on each, and times split into your hands and the assistant's. Start with the placement interview, then one correction-only chat.

Two questions decide every prompt: does the whole job fit in one context, and can you afford to run it again tomorrow. Those two answers give you the split pick.

1. The 60-Second Placement Interview

Paste it into either assistant and answer six questions out loud. Each answer decides the next question, so question four is harder than question one.

You get a level from A1 to C2, three strengths, three gaps and one 15-minute task to start on. No grammar test, no exam.

Both assistants place people a little high from six answers. If the level feels flattering, drop it one step before you use it.

Claude Free can also stop halfway, because its rolling five-hour message limit can land between two questions. ChatGPT Free runs text chats without a cap that stops you this way.

By hand this placement takes about 8 minutes. The interview itself runs about 30 seconds, and your review of the level and gaps takes 4 minutes.

Claude holds all six questions in one long context and keeps its $0 tier ad-free, so it takes this one; ChatGPT Free finishes an interrupted interview in a second chat.

2. The Correction-Only Conversation Partner

Yes, with one setup. Tell the assistant it talks for 10 turns and says nothing about mistakes until the end.

It then hands back three errors that changed what you meant, two phrases worth reusing, one pattern to drill and one answer to redo. The whole run takes about 12 minutes and 40 seconds of AI reply.

27K tokens. A token is a small chunk of text, roughly three-quarters of a word. That's the context window, the total amount the chat can hold and reread. At 27K it fits about 12 pages, while Claude holds 200,000 tokens.

So the bill looks like this. ChatGPT Free carries a weekly usage limit. Claude adds a rolling five-hour session limit that can cut a ten-turn conversation early. Claude Free runs with no ads.

ChatGPT Free shows ads in Germany as of October 2026, and Go at $8 carries them, while the $20 tier and above do not.

For daily volume, ChatGPT Free wins. For one long, ad-free conversation and a full report on the errors, Claude Pro at $20 a month, or $17 a month billed annually, earns the $20 over Go.

Review and the redo take about 5 minutes; save the two phrases as flashcards afterward.

3. The Real-Life Role-Play Rehearsal

This prompt builds a speaking rehearsal with a scene, a character and a limit. A pharmacy counter, a hotel check-in, a job interview: each one asks a question the general conversation never reaches.

You set the role, the place and a fixed run of 8 exchanges, and the assistant stays in character while you answer.

Both assistants start to sound less like the role as the scene goes long. The pharmacist turns into a general tutor by exchange six, and the stakes disappear.

The check is the transcript: if it stops reading like a counter conversation, the character has slipped. Claude can hold longer scenes if you paste in a detailed scenario brief first.

ChatGPT Free is better for a four-exchange warm-up, where the drift never gets a chance to start.

By hand this takes about 10 minutes, the AI plays the scene in about 30 seconds, and the debrief takes 4 minutes to read.

Replay the three better lines out loud, then run the scene again.

4. The Grammar Rule in Context

The catch first: both assistants will state a Spanish grammar rule with total confidence and get it wrong. So check the rule against a dictionary or grammar source before you memorize it.

The safe use: paste one sentence you actually wrote, the meaning you wanted, and let the tool name the single biggest problem in under 120 words.

What you get: two correct examples, one common trap, and a new sentence to write. Six minutes by hand, about 20 seconds for the answer, 3 minutes to check and write.

Claude explains it more clearly and ad-free; ChatGPT Free handles a 20-second rule check. Add both sentences to the error log.

5. The Article-to-Vocabulary Mining Prompt

Fifteen minutes of hunting a dictionary for words you might never reuse. An article you actually want to read has the vocabulary your next conversation needs, and it takes about 45 seconds to turn into cards.

The prompt: paste it with the article, then read the twelve entries. Each gives the word, its part of speech, a short English definition, the sentence from the article, and one new sentence at your level. Six fill-in-the-blank questions follow.

The same job, each tool: in Claude, paste a whole chapter or PDF and get twelve cards back, because the 200,000-token window holds the piece and Free takes up to 20 files per chat.

ChatGPT Free splits anything past roughly 12 pages, and its file Library caps at 500MB.

Claude: the pick for long articles, chapters and PDFs. ChatGPT Free: fine for a one-page news story. Either way, 5 minutes of review, then delete the four words nobody would ever say.

6. The Nuance Check for Translations

The slow part is not writing the translation. It is staring at a sentence that is grammatically perfect and knowing, somehow, that no native speaker would say it.

Line your version up against the source and the tool returns a natural version, a literal version, the two differences that matter, one note on formal versus informal, and a two-choice drill, the kind of prompt trick that repairs weak answers. Twenty-five seconds for the answer, four minutes to read, choose and file.

Both assistants invent idiomatic usage, so check a phrase with a native speaker before you use it out loud.

Claude holds the nuance across a longer explanation; ChatGPT Free is enough for one line. The cost: eight minutes by hand, and $0 on either free tier.

7. The Red-Pen Writing Feedback Prompt

A correction you can apply yourself beats a clean rewrite you only admire.

Pasted into Claude, a short Spanish paragraph came back with three clarity errors, one rule for each, a bolded corrected copy and a register note, then a line asking for the learner's own second draft.

Both quiet down and rewrite the whole thing unless the no-rewrite line is repeated twice, so check the first reply for your own sentences.

Claude holds that instruction across a long essay; ChatGPT Free handles one paragraph fine. About 35 seconds for the feedback, 12 minutes by hand, 6 minutes to rewrite, then compare the two.

8. The Mini-Comprehension Test Builder

One 150-word Spanish passage at a fixed CEFR level, followed by five questions and no answers until you reply, separates the two.

ChatGPT Free is the tool for this, because its unlimited text chats let you run round after round.

Step 1 – set the test. Paste the prompt below into a fresh chat. Say B1, Spanish, and a work-and-travel topic. The reply should be one passage of roughly 150 words, then the first question alone.

Step 2 – answer, then read the grade. Reply to all five, then send. Expect each miss explained in a short English sentence and one line named as the sentence to re-read.

Step 3 – check the level. If the passage reads far above B1, say so and ask for a shorter one with simpler words. Step 4 – re-read.

Go back to that one sentence, 5 minutes in all after about 40 seconds of generation. Claude gives the cleaner explanation of a miss, without ads. Budget 10 minutes by hand.

9. The Two-Week Study Plan Prompt

That single paste is a one-time setup that repays itself for a fortnight. It returns fourteen dated days, each with a core task and a short review, a checkpoint at the end of each week, and no paid material. Filling it takes about 50 seconds.

Claude wins it, because the whole plan stays inside one long context and comes back consistent from day one to day fourteen, on the ad-free free tier. ChatGPT Free is fine if you build one week at a time.

The catch: both overstuff a plan, so the daily minute count is the rule, not a suggestion. Claude Free's rolling five-hour session limit can also split the build across two sittings, so start it fresh.

Review and trim in 8 minutes against the 20 it would take on your own, then print only the 10-minute core tasks and pin them where you work.

10. The Slang-and-Culture Reality Check

Five minutes of checking, about 25 seconds of it the assistant's answer, and 3 minutes yours: that is the real price of a phrase.

The 14-day plan allots minutes, but no schedule tells you whether "me da igual" lands as relaxed or rude in a work-and-travel chat.

Paste the phrase you intend to use into Claude for a longer, ad-free reading, or ChatGPT Free when you want a current web check.

Both assistants invent slang and regional usage, and an invented source looks exactly like a real one.

Treat any unsourced claim as a guess and run it past a native speaker. Cost of the check: 5 minutes and $0, then try the safer alternative in your next conversation.

11. The Exam-Interview Simulation Prompt

The dull part of exam prep is rehearsing alone: no clock, no examiner waiting on your answer, no score at the end.

An exam simulation handles it if you name the exam, the level and the four scores you want, and accept that pronunciation is only an estimate in text.

Both free tiers can run it for $0.

The catch is the pronunciation score. Typed answers tell the assistant nothing about how you sound, so switch to voice mode and treat every mark as rehearsal, never an official result.

ChatGPT Free handles repeated timed runs best, since text and voice chats are both there. Claude Free has voice mode but a tighter message window.

By hand the mock runs 15 minutes; the AI takes about 60 seconds, and reviewing plus the retry takes 6.

Verdict: ChatGPT Free for repeat runs under the clock. Verdict: Claude Free for one careful, ad-free run.

Frequently Asked Questions

Do I need to pay for either assistant to use these prompts?

No. Both Claude Free and ChatGPT Free run every prompt here. Paid plans matter for volume and context: ChatGPT Go at $8 a month adds headroom, ChatGPT Plus at $20 and Claude Pro at $20 – or $17 a month billed annually – remove many free-tier limits.

Can I paste a private document or client file into these chats?

No. Never paste passwords, client files, or medical or financial records into either assistant. On free and personal plans, chats may be used to train models unless you turn that off in settings, so keep practice material separate from real work.

Which assistant should I pick for a long book chapter or PDF?

Claude. Its non-Enterprise context window is 200,000 tokens, and Claude Free allows up to 20 files per chat at 500MB each. ChatGPT Free has a 27K context window and about a 12-page input maximum, so long documents need splitting.

How accurate is the pronunciation score?

Not very, if you are typing. A text chat cannot judge pronunciation reliably. Use voice mode and treat the score as practice, not an official result, then have a teacher or native speaker confirm the sounds you are unsure about.

Why does my level come back higher than I expected?

Both assistants overestimate from a six-answer sample. If the first practice task feels hard, drop one level band and rerun the placement interview in a new chat with a note saying you found the task difficult.

Pick Your Tutor by How You Practice

Claude is the surprising overall pick for long-context, correction-heavy and document-based tutoring. It holds a 200,000-token window, keeps a scenario brief in character, and its free tier is ad-free.

ChatGPT Free or Go is the better daily driver: unlimited text chats, quick voice drills and a cheaper step up at $8.

Start with the placement interview and one correction-only conversation. Run those two, note which assistant finished without hitting a wall, and build the rest of your practice around that one.