Free TOEFL iBT 2026 prep — Telegram bot

The 4 sections of TOEFL iBT 2026

The key novelty in the 2026 redesign is adaptive testing in Reading and Listening: question difficulty adjusts to your performance in real time, giving a more accurate picture of your level. Reading and Listening no longer have "everyone gets the same questions".

The 1-6 scale — what each score means

• 6 — Advanced. Near-native English in academic use. Approx. CEFR C2.
• 5 — High-Intermediate / Advanced. Approx. CEFR C1. Common minimum for top US graduate programs.
• 4 — Intermediate. Approx. CEFR B2. Common minimum for many US universities.
• 3 — Low-Intermediate. Approx. CEFR B1. Some universities accept with English support programs.
• 2 — Basic. Insufficient for academic use.
• 1 — Below Basic.

Each candidate gets a section score plus total. Universities set their minimum per section (some require ≥4 in Speaking even if total is higher).

Listen and Repeat — the new Speaking task

This is the biggest difference from old TOEFL. You hear a sentence by audio ONCE and must repeat it exactly in 8 to 12 seconds. There are 7 sentences per test, each 5 to 14 words long, on campus or daily life topics.

It's NOT a creativity test. AI measures:

• Intelligible pronunciation — can it be understood?
• Natural rhythm — does the flow mimic native English?
• Correct stress — do you emphasize the right syllables?
• Short-term memory — you hear it once, no replay.

The bot replicates it faithfully: TTS reads the sentence (Gemini native voice), you send a voice message, Whisper transcribes, AI evaluates against the target. Feedback in seconds.

Build a Sentence — the new Writing task

10 sentences per test. Each comes with scrambled words (sometimes with one extra distractor that's NOT used). You arrange them. 1 point if perfect, 0 if there's ANY single error in order or extra/missing word. You have ~7 minutes for all 10.

The bot replicates this with tap-to-arrange buttons in Telegram: you tap words in your believed order, the bot validates exact-match against the target sentence. No AI needed — either the order is exact or it isn't.

Speaking — the 4 official ETS dimensions

Since 2026, ETS scores Speaking entirely with AI. Each response is evaluated on 4 dimensions:

• Fluency — flow, pauses, appropriate speed.
• Intelligibility — clear pronunciation for native speakers.
• Language Use — grammatical and lexical accuracy.
• Organization — coherent structure, idea connection.

The bot calibrates its AI rubric to those exact 4 dimensions. Not generic feedback: feedback aligned to the real scoring method.

How to start

1. Open the bot: t.me/PreparaTOEFLBot
2. Tap /start. It asks for target score and weak section.
3. Get 3–5 daily questions automatically at your preferred hour.
4. Practice Listen and Repeat, Build a Sentence, Speaking, and Writing whenever with AI feedback.

📲 Open Telegram bot

Other bots and resources

• AptisBot — main page (English)
• IELTS — independent bot Academic + General
• Aptis vs IELTS vs TOEFL — comparison
• Versión en español de esta página