PTE 2025 Scoring Changes Explained: What Changed, What It Means for You
On 7 August 2025 Pearson overhauled PTE Academic: hybrid AI-plus-human scoring, template detection penalties, two new task types, and new per-skill score thresholds for Australia. Here is exactly what changed, what stayed the same, and how to prepare.
Published 20 June 2026 · 14 min read · PTE Mocks editorial team
In one line
On 7 August 2025, Pearson launched a significantly updated PTE Academic. The headline: a hybrid scoring model (AI plus trained human reviewers), template detection that penalises memorised scripts, two new task types, expanded scoring rubrics, and new per-skill score thresholds for Australian, New Zealand and UK migration. If you are preparing for PTE in 2026, every one of these changes affects your strategy. Last reviewed 20 June 2026.
What changed on 7 August 2025: the full picture
Pearson describes the update as an “enhanced” PTE Academic, but it is more than a tweak. Five interconnected changes went live at once, representing the biggest overhaul to the PTE scoring system since the test launched in 2009:
| Change | What it does | Who it affects most |
|---|---|---|
| Hybrid scoring | Seven task types now get a human content review on top of AI scoring | All test takers, especially those who relied on vague or off-topic answers |
| Template detection | AI flags and penalises memorised, robotic answers by up to 30% | Students using memorised scripts for Speaking and Writing |
| Two new task types | Respond to a Situation and Summarize Group Discussion join the test | All test takers (no one has practised these before) |
| Expanded scoring rubrics | Content scoring scales widened (e.g. Describe Image content from 0-5 to 0-6) | All test takers, especially those scoring in the 65-79 range |
| New score thresholds | Australia, NZ and UK updated per-skill PTE requirements | Migration applicants targeting Proficient or Superior English |
Each of these is covered in detail below, with what stayed the same, so you know exactly what to adjust. (Source: Pearson PTE changes 2025, Gurully PTE scoring analysis.)
Hybrid scoring: AI plus human review (the biggest shift)
Before August 2025, PTE Academic was scored entirely by AI. Now, seven task types use “double marking”: the AI scores your response, a trained human examiner independently reviews the content, and if the two disagree by more than a threshold, a second human makes the final call.
Why this matters
This is the most significant change to PTE scoring since the test was created. Previously, students could score well by “sounding fluent” without delivering relevant content. The human review layer now catches this gap.
The seven tasks under hybrid scoring are:
| Task | Section | What AI scores | What the human reviews |
|---|---|---|---|
| Describe Image | Speaking & Writing | Pronunciation, oral fluency | Content accuracy, relevance to the specific image |
| Re-tell Lecture | Speaking & Writing | Pronunciation, oral fluency | Content coverage of the lecture's key points |
| Respond to a Situation (new) | Speaking & Writing | Pronunciation, oral fluency | Appropriacy, tone, relevance to the scenario |
| Summarize Written Text | Speaking & Writing | Grammar, vocabulary | Content relevance, key point coverage |
| Write Essay | Speaking & Writing | Grammar, vocabulary, spelling | Development, structure, coherence, linguistic range |
| Summarize Spoken Text | Listening | Grammar, vocabulary | Content accuracy, key point coverage |
| Summarize Group Discussion (new) | Listening | Grammar, vocabulary, form | Content relevance, synthesis of multiple viewpoints |
Source: Pearson PTE Academic scoring, PTE Magic scoring analysis.
Crucially, the examiner never sees your name, photo or nationality, so there is no accent or identity bias in the content review. Pronunciation, oral fluency, grammar and spelling remain AI-only. The human layer is purely about whether your content addresses the prompt.
Expanded scoring rubrics: before vs after
Beyond adding human review, Pearson also widened the scoring scales for several tasks. This is a detail many students miss, but it directly affects how points are distributed. A wider scale gives examiners (both human and AI) more granularity to differentiate between good and excellent answers.
| Task | Criterion | Old scale | New scale | Impact |
|---|---|---|---|---|
| Describe Image | Content | 0-5 | 0-6 | More differentiation at higher levels |
| Re-tell Lecture | Content | 0-3 | 0-6 | Content now weighted much more heavily |
| Summarize Written Text | Content | 0-2 | 0-4 | Double the content weight |
| Write Essay | Development, structure, coherence | 0-2 | 0-6 | Major increase in essay depth expectations |
| Write Essay | General linguistic range | 0-2 | 0-6 | Vocabulary sophistication matters more |
Source: Gurully scoring analysis, cross-referenced with Pearson's official PTE Academic Score Guide (PDF).
Practical takeaway
The Re-tell Lecture content scale jumped from 0-3 to 0-6, effectively doubling the weight of what you actually say versus how you say it. A fluent but vague retelling that scored 2/3 on the old scale might only score 2/6 on the new one. You need to capture the lecture's specific key points, not just paraphrase generally.
Question count changes per task
The number of questions per task type also shifted in the 2025 update. Some tasks increased in count (giving them more weight in your score), while others decreased:
| Task | Old count | New count | Change |
|---|---|---|---|
| Describe Image | 3-4 questions | 5-6 questions | +2 questions |
| Re-tell Lecture | 1-2 questions | 2-3 questions | +1 question |
| Summarize Written Text | 2 questions | 1 question | -1 question |
| Respond to a Situation | Did not exist | 2-3 questions | New |
| Summarize Group Discussion | Did not exist | 1-2 questions | New |
The increase in Describe Image questions is particularly significant. With 5-6 questions instead of 3-4, this task now carries more weight in your Speaking and Writing score. Students who struggle with Describe Image will feel the effect more than before. The decrease in Summarize Written Text (from 2 to 1) means each SWT response matters more individually. You get one shot instead of two.
Template detection: what it catches and how much it costs
This is the change that has affected the most students. Pearson's updated AI now includes template detection that identifies memorised, formulaic responses and penalises them. E2Language describes this shift as: “Authentic communication has officially replaced the magic template.”
What the system looks for:
- Rhythmic monotony: the flat, robotic cadence of a recited script, without natural pauses, stress variation or self-correction.
- Formulaic structure: answers that follow a rigid template regardless of the actual prompt (the same opening sentence, the same linking phrases in the same order, the same conclusion).
- Content mismatch: answers that sound polished but do not address the specific image, lecture or essay prompt.
Score impact
Community reports and coaching forums suggest template-flagged responses can lose up to 30% of their content score, which translates to a 10 to 15 point drop on the real test compared to mock scores where no such detection exists. This is the single most common reason students score significantly lower on the real PTE than on third-party mocks.
Structured approaches are fine; memorised scripts are not. Having a framework for Describe Image (“The image shows... The main trend is... In conclusion...”) is good strategy. Reciting an identical 40-second answer with the same phrases regardless of what the image actually shows is what gets flagged. The distinction is whether your specific content changes with each prompt.
Here is how to tell the difference:
| Approach | Example | Detection risk |
|---|---|---|
| Structured framework (safe) | “The image shows [specific content]. The main trend is [specific observation]. In conclusion, [specific summary].” | Low. The skeleton is fixed, but the content changes every time. |
| Memorised template (risky) | “The given image illustrates valuable information regarding the topic. There are several noteworthy points to mention. Firstly... Secondly... In conclusion, it can be summarized that...” | High. Same filler phrases regardless of prompt. Flagged for rhythmic monotony and content mismatch. |
Two new task types: Respond to a Situation and Summarize Group Discussion
The Speaking and Writing section now includes 22 task types (up from 20), with two additions designed to test spontaneous, real-world English:
| New task | What it tests | Format | Scoring criteria |
|---|---|---|---|
| Respond to a Situation | Pragmatic, real-world speaking (making a request, explaining a problem, giving advice) | You read a short scenario, then record a 40-second spoken response | Content, appropriacy, oral fluency, pronunciation (0-6 each) |
| Summarize Group Discussion | Listening comprehension and written synthesis of a 3-way conversation with conflicting viewpoints | You listen to a multi-speaker discussion (about 120 seconds), then write a summary | Content, form, grammar, vocabulary, spelling |
These tasks reward natural, spontaneous English over rehearsed scripts, which fits the broader direction of the 2025 update. Respond to a Situation in particular cannot be templated because the scenario changes every time and the appropriate response depends entirely on the specific context (tone, audience, purpose).
Section timing change
The Speaking and Writing section is now 76 to 84 minutes (up from around 54 to 67 minutes before). This is partly because of the new tasks and partly because Summarize Written Text was reduced from two questions to one. Plan your pacing accordingly.
Source: E2Language PTE 2026 changes, Pearson PTE.
New per-skill score thresholds for Australia PR
Before August 2025, Australia used flat PTE thresholds: Competent was 50 in every skill, Proficient was 65, and Superior was 79. The new test uses different thresholds per skill, which makes some levels easier and others harder to reach:
| Level (points) | Listening | Reading | Writing | Speaking |
|---|---|---|---|---|
| Competent (0 pts) | 47 | 48 | 51 | 54 |
| Proficient (10 pts) | 58 | 59 | 69 | 76 |
| Superior (20 pts) | 69 | 70 | 85 | 88 |
Source: Australian Department of Home Affairs English language requirements. Tests taken on or before 6 August 2025 use the old flat thresholds until 6 August 2028.
The Superior challenge
Superior English now demands Writing 85 and Speaking 88. Compare that with Listening 69 and Reading 70. The gap is enormous. If you are chasing the 20-point Superior band, Writing and Speaking are where you need to concentrate. For many students, the old “PTE is easier than IELTS for Superior” no longer holds.
Our PTE vs IELTS for Australia PR guide covers this comparison in detail.
New Zealand and UK score requirements
Australia is not the only country that updated its PTE requirements. New Zealand and the United Kingdom also adopted new per-skill thresholds aligned with the updated test:
New Zealand updated its Skilled Migrant Category and other visa pathways to reflect the new PTE scoring. Immigration New Zealand (INZ) requires PTE Academic scores that map to specific CEFR levels, and the updated test produces scores calibrated against the revised rubrics.
United Kingdom continues to accept PTE Academic for Skilled Worker, Student and other visa routes. UKVI-approved PTE tests (PTE Academic UKVI) now follow the same hybrid scoring model, with the same template detection and human review. The score requirements for each visa type remain set by the Home Office, but the underlying scoring has changed.
Key point for NZ and UK applicants
Even if the threshold numbers for your visa type have not changed, the way those scores are calculated has. The hybrid scoring model means your score may be slightly different from what it would have been under the old pure-AI system. Aim for a comfortable margin above the requirement.
Source: Pearson PTE changes 2025, Immigration New Zealand, UK Home Office guidance.
What stayed the same (and why it matters)
Not everything changed. Knowing what is the same helps you focus your preparation adjustments on what actually shifted:
| Aspect | Status | Detail |
|---|---|---|
| Score scale | Unchanged | Still 10 to 90, still maps to CEFR and GSE |
| Test length | Unchanged | Still about two hours total |
| Delivery format | Unchanged | Still fully computer-based at a Pearson test centre |
| Results speed | Unchanged | Still typically about 48 hours |
| Score validity | Unchanged | Still two years |
| Core task types | Unchanged | Read Aloud, Repeat Sentence, Write from Dictation, Fill in the Blanks, Multiple Choice, Re-order Paragraphs, and all Reading and Listening items |
| Pronunciation/fluency scoring | Unchanged | Still AI-only, still accent-neutral |
| Integrated scoring | Unchanged | Tasks still feed multiple communicative skills (Read Aloud scores both Reading and Speaking) |
| Global acceptance | Unchanged | Still accepted by 3,500+ institutions globally |
Source: Pearson PTE Academic scoring, E2Language.
If you took PTE before August 2025
Tests taken on or before 6 August 2025 are treated as the legacy PTE Academic. Your existing score report remains valid, and the old flat thresholds (50 / 65 / 79 for Competent / Proficient / Superior in Australia) still apply to those results. Australia will accept legacy scores as evidence until 6 August 2028.
If you are retaking PTE now, you sit the new format and your score is assessed against the new per-skill thresholds. You cannot choose the old format.
Strategic note
If you have a legacy PTE score that meets your requirements, there is no reason to retake the test unless it expires before your application is processed. Legacy scores are fully valid until 6 August 2028.
How to prepare for the new PTE scoring system
The changes reward genuine English ability over test tricks. Here is what to adjust in your preparation strategy:
- Stop using memorised templates. A structured approach is fine (have a framework for essays and Describe Image), but the specific content must change with every prompt. If you have been reciting identical scripts, stop now. Templates are the single biggest score risk in the 2025 format.
- Practise spontaneous speaking. Respond to a Situation cannot be templated. Practise speaking naturally about real-world scenarios: requesting, explaining, advising, persuading. The goal is fluent, relevant content, not polished recitation.
- Focus on content relevance. With human reviewers now checking content, generic or off-topic answers are caught more reliably. Read or listen to the prompt carefully and make sure your answer addresses it specifically.
- Invest more time in Describe Image. With 5-6 questions instead of 3-4, and content scored on a wider 0-6 scale, this task carries significantly more weight than before. Practise describing graphs, charts, maps and diagrams with specific data points.
- Target your weakest skill. The new per-skill thresholds mean you cannot coast on strong Reading and Listening to compensate for weak Writing and Speaking. Each skill has its own bar, and your lowest skill caps your English level. Take a full mock test to identify the bottleneck.
- Use the enabling skill breakdown. A mock score of 72 is not useful on its own. The enabling skill breakdown (pronunciation, oral fluency, grammar, vocabulary, written discourse, spelling) tells you exactly what to fix. Our plateau guides map common bottlenecks by score range.
How the 2025 changes affect mock test scores
Third-party mock tests (including ours) use different AI models than Pearson, and none of them include human content review or template detection. Understanding this gap is critical for accurate preparation:
| Factor | Mock tests | Real PTE (post-August 2025) | Score impact |
|---|---|---|---|
| Template detection | Not present | Active, penalises up to 30% | Mocks inflate scores for template users by 10-15 points |
| Human content review | Not present | Active on 7 task types | Mocks may miss off-topic or vague content |
| Scoring rubrics | Approximated from Pearson criteria | Official expanded scales (0-6) | Minor variance, typically 3-7 points |
| Improvement trends | Reliable | Reliable | Trend is consistent even if absolutes differ |
The bottom line: use mocks for tracking improvement and identifying weak enabling skills, not as an exact score prediction. If your mock scores go from 62 to 67 to 71, the improvement is real. But if you are scoring well on mocks using memorised templates, expect a significant drop on the real PTE.
For a deeper analysis, see our guides on mock test accuracy and why scores differ.
Timeline of PTE scoring changes
| Date | What happened |
|---|---|
| 2024 | Pearson pre-announces the changes and begins pilot testing of the hybrid scoring model with select test centres. |
| November 2024 | Double scoring (AI + human review) implementation begins in pilot markets. (Source: PTE Magic.) |
| 7 August 2025 | New PTE Academic goes live globally: hybrid scoring, template detection, Respond to a Situation, Summarize Group Discussion. New per-skill score thresholds for Australia, NZ and UK take effect. (Source: Pearson.) |
| 6 August 2028 | Deadline for Australian migration to accept legacy PTE scores under the old flat thresholds (50/65/79). |
Frequently asked
What changed in PTE Academic in August 2025?
Five things: hybrid AI-plus-human scoring on seven task types, template detection that penalises memorised answers, two new task types (Respond to a Situation and Summarize Group Discussion), expanded scoring rubrics with wider scales, and new per-skill score thresholds for Australia, New Zealand and the UK.
Is PTE still scored by AI after the 2025 changes?
Yes, but not AI alone. Seven task types now use hybrid scoring: AI scores the response, a human reviews the content, and if they disagree a second human decides. Pronunciation and fluency remain AI-only.
Do templates still work in PTE 2025?
No. Pearson's updated scoring model includes template detection that penalises memorised, formulaic answers. Community reports suggest template-flagged responses can lose up to 30% of their content score. Structured approaches (a framework where specific content changes each time) are fine; identical recited scripts are penalised.
What are the new PTE scores for Australia PR?
Since 7 August 2025, Australia uses per-skill thresholds instead of a flat score. Competent: L47 R48 W51 S54. Proficient (10 pts): L58 R59 W69 S76. Superior (20 pts): L69 R70 W85 S88. Tests taken before 7 August 2025 use the old flat thresholds until 6 August 2028.
Is PTE harder after August 2025?
It is harder for students who relied on memorised templates, because those are now detected and penalised. For students with genuine English ability who prepare with real practice, the difficulty is similar. The new task types test natural communication, not academic tricks.
What is Respond to a Situation in PTE?
A new Speaking task where you read a short real-world scenario (a complaint, a request, a workplace situation) and record a 40-second spoken response. It tests pragmatic English (appropriate tone, register, and content) and cannot be templated because the scenario changes every time.
What is Summarize Group Discussion in PTE?
A new task where you listen to a discussion between multiple speakers (about 120 seconds) and write a summary synthesizing the different viewpoints. It tests listening comprehension and written summary skills, and is one of the seven tasks under hybrid scoring.
How did the scoring rubrics change in PTE 2025?
Several tasks got wider scoring scales. Describe Image content went from 0-5 to 0-6, Re-tell Lecture content from 0-3 to 0-6, Summarize Written Text content from 0-2 to 0-4, and Write Essay criteria from 0-2 to 0-6. Wider scales give examiners more granularity to differentiate between good and excellent answers.
Do my old PTE scores still count?
Yes. Tests taken on or before 6 August 2025 are treated as legacy PTE Academic. For Australian migration, the old flat thresholds (50 / 65 / 79 for Competent / Proficient / Superior) apply to those results until 6 August 2028.
How do mock test scores compare to the real PTE after 2025?
Third-party mocks do not have human content review or template detection, so template-heavy answers score higher on mocks than on the real test. Typical variance is 3-7 points for non-template users. Use mocks for tracking improvement trends and identifying weak enabling skills, not as an exact score prediction.
Should I still take mock tests after the 2025 changes?
Yes. Mocks remain the best preparation tool for identifying weak skills, building exam stamina, and tracking improvement. Just avoid relying on memorised templates that inflate mock scores but get penalised on the real test.
How many questions are in Describe Image now?
Describe Image increased from 3-4 questions to 5-6 questions in the 2025 update. Combined with the wider content scoring scale (0-6 instead of 0-5), this task now carries significantly more weight in your Speaking and Writing score.
Put it to the test
Free, full-length PTE mock tests, scored by AI. See where you really stand.