JEE Advanced · 2025 and 2026
Can AI solve JEE?
Quick answer
Yes. On the official JEE Advanced 2026 papers, scored blind against the IIT key, Claude Fable 5.1 scored 358 out of 360 (99.4%). GPT 6 Astra scored 356, GPT 5.5 scored 351, and Claude Sonnet 5 scored 341. On JEE Advanced 2025, Claude Fable 5.1 scored a clean 360/360.
People asking whether AI can solve JEE usually mean JEE Advanced — two official papers, 360 marks, Physics, Chemistry, and Mathematics. Goodmarks runs named frontier models on those printed English booklets. The model sees the question image only. The official key is applied after it replies.
JEE Advanced 2026 AI scores
Held May 17, 2026. 102 questions across Paper 1 and Paper 2. Marks out of 360.
| Model | Provider | Score | Percent | Tally |
|---|---|---|---|---|
| Claude Fable 5.1 | Anthropic | 358/360 | 99.4% | 101 correct · 1 partial |
| GPT 6 Astra | OpenAI | 356/360 | 98.9% | 101 correct · 1 wrong |
| GPT 5.5 | OpenAI | 351/360 | 97.5% | 100 correct · 2 wrong |
| Claude Sonnet 5 | Anthropic | 341/360 | 94.7% | 96 correct · 2 partial · 2 wrong · 2 error |
JEE Advanced 2025 AI scores
Held May 18, 2025. 96 questions. Same 360-mark official key.
| Model | Provider | Score | Percent |
|---|---|---|---|
| Claude Fable 5.1 | Anthropic | 360/360 | 100% |
| GPT 6 Astra | OpenAI | 352/360 | 97.8% |
| GPT 5.6 Luna | OpenAI | 319/360 | 88.6% |
How Goodmarks scores AI on JEE
- Split the official papers. Cut the English Paper 1 and Paper 2 booklets into single-question images. Options, figures, and matching lists stay as printed.
- Keep every model blind. Send only the question image and its type. The official key never enters the prompt.
- Ask for a structured answer. Each model can think first. Scoring uses the final JSON answer, not the prose around it.
- Mark with the IIT key. Apply official solutions after the reply, including numerical ranges and multi-correct partial marks.
JEE Main vs JEE Advanced
These results are JEE Advanced only. JEE Main is a different exam with a larger candidate pool and a different paper pattern. We have not published a JEE Main AI leaderboard yet. If you are practising for JEE Main, use the year-wise JEE Main hubs and online practice.
See every question and model
-
Full JEE Advanced leaderboard
2025 and 2026 scores, paper and subject splits, and every miss.
-
ChatGPT JEE Advanced score
GPT 6 Astra, GPT 5.5, and GPT 5.6 Luna on official papers.
-
Claude JEE Advanced score
Fable 5.1 at 358/360 in 2026 and 360/360 in 2025.
-
AI cracks JEE Advanced 2026
Four-model write-up and the Paper 2 Mathematics question 4 rescore.
-
GPT-5.5 question log
Every printed question, official key, and GPT-5.5 reasoning.
-
Practise JEE yourself
Syllabus-aligned MCQs in Physics, Chemistry, and Mathematics.
Frequently asked questions
Can AI solve JEE?
Yes. On official JEE Advanced 2026 papers, scored blind against the IIT key, Claude Fable 5.1 scored 358 out of 360 (99.4%). GPT 6 Astra scored 356/360. GPT 5.5 scored 351/360. Claude Sonnet 5 scored 341/360. On JEE Advanced 2025, Claude Fable 5.1 scored 360 out of 360 (100%).
Can AI solve JEE Advanced 2026?
Yes. Four frontier models sat both official English papers without seeing the answer key. Claude Fable 5.1 leads at 358/360 (99.4%). All four models scored above 94%.
How does AI perform on JEE?
On JEE Advanced — the harder of the two exams — the latest Goodmarks run puts models between 341 and 358 out of 360 in 2026, and between 319 and 360 out of 360 in 2025. We do not yet publish a JEE Main AI leaderboard.
Can ChatGPT solve JEE Advanced?
OpenAI models on the 2026 official papers scored GPT 6 Astra 356/360, GPT 5.5 351/360. GPT 5.5 shipped on 23 April 2026; JEE Advanced 2026 was held on 17 May 2026, so that paper did not exist when GPT 5.5 went public.
Can Claude solve JEE Advanced?
Anthropic models on the 2026 official papers scored Claude Fable 5.1 358/360, Claude Sonnet 5 341/360. Claude Fable 5.1 also scored a clean 360/360 on JEE Advanced 2025.
What is ChatGPT’s JEE score?
On official JEE Advanced 2026, GPT 6 Astra scored 356/360 and GPT 5.5 scored 351/360. On JEE Advanced 2025, GPT 6 Astra scored 352/360 and GPT 5.6 Luna scored 319/360.
Did the models see the official JEE answers?
No. Each request included only the printed question image and its type (single correct, multi correct, numerical, or matching). Official IIT keys were applied only after the model replied.
Is this JEE Main or JEE Advanced?
These scores are JEE Advanced only — Paper 1 and Paper 2 of the official English booklets, 360 marks. JEE Main is a different exam. Goodmarks has not published a JEE Main AI benchmark yet.