HiddenMath
DiffusionGemma 26B A4B 80.6% leads since Jul 2026 Dropped Models solve 179 competition-level math problems that experts wrote from scratch. Answers are graded automatically.
In-house Google DeepMind set that was never released.
Made by Google DeepMind Released May 2024 Competition problems with answers Tasks kept private Last in a launch post Jun 2025 , dropped before it saturated
Last model added Jul 2026
- Level
- Competition The strongest students, in olympiad-style contests
- Scored
- Checked automatically A program checks the result against an answer key, unit tests or the expected final state.
- Answers
- Short answer A number, word or short string matched to a key.
11
Models scored in lab cards
7
Launch posts list it, from 1 lab
7
System cards report it, from 1 lab
DiffusionGemma 26B A4B leads with 80.6% among models released by Jun 2026
Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.
The newest model with a score here came out in Jun 2026, and nobody has reported a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6 Astra
- Claude Opus 5.5
- GPT-6.1 Terra
- GPT-6.1 Sol
- and 16 more
- 1 DiffusionGemma 26B A4B 80.6%
- 2 Gemini 2.0 Pro 65.2%
- 3 Gemini 2.0 Flash Feb 2025 63.5%
- 4 Gemini 2.0 Flash Dec 2024 63%
- 5 Gemma 3 27B 60.3%
- 6 Gemini 2.0 Flash-Lite 55.3%
- 7 Gemma 3 12B 54.5%
- 8 Gemma 3 4B 43%
- 9 Gemma 3n E4B 37.7%
- 10 Gemini 1.5 Pro May 2024 20.1%
- 11 Gemma 3 1B 15.8%
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | DiffusionGemma 26B A4B | 80.6% | Jul 2026 | Model card | |
| 2 | Gemini 2.0 Pro | 65.2% | Feb 2025 | Model card | |
| 3 | Gemini 2.0 Flash Feb 2025 | 63.5% | Feb 2025 | Model card | |
| 4 | Gemini 2.0 Flash Dec 2024 | 63% | Dec 2024 | Model card | |
| 5 | Gemma 3 27B | 60.3% | Mar 2025 | Model card | |
| 6 | Gemini 2.0 Flash-Lite | 55.3% | Feb 2025 | Model card | |
| 7 | Gemma 3 12B | 54.5% | Mar 2025 | Model card | |
| 8 | Gemma 3 4B | 43% | Mar 2025 | Model card | |
| 9 | Gemma 3n E4B | 37.7% | Jun 2025 | Model card | |
| 10 | Gemini 1.5 Pro May 2024 | 20.1% | May 2024 | Model card | |
| 11 | Gemma 3 1B | 15.8% | Mar 2025 | Model card |
4 models have held number one since its release in May 2024
Lab claims count from the day the card came out, independent runs from the model’s release.
Google DeepMind
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Reported by the lab in its model card
Best so far
No launch post has listed it since Jun 2025
Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.
| Date | Launch post | Score | As the post writes it |
|---|---|---|---|
| 26 Jun 2025 | Gemma 3n E4B | 37.7 | HiddenMath 0-shot (IT) |
| 12 Mar 2025 | Gemma 3 27B | 60.3 | HiddenMath 0-shot (IT) |
| 5 Feb 2025 | Gemini 2.0 Flash-Lite | 55.3 | HiddenMath |
| 5 Feb 2025 | Gemini 2.0 Flash | 63.5 | HiddenMath |
| 5 Feb 2025 | Gemini 2.0 Pro | 65.2 | HiddenMath |
| 11 Dec 2024 | Gemini 2.0 Flash | 63 | HiddenMath |
| 14 May 2024 | Gemini 1.5 Pro | 20.1 | HiddenMath, 0-shot |
7 system cards report it, from 1 lab
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| 31 Jul 2026 | DiffusionGemma 26B A4B technical report | 80.6 | HiddenMath (internal), TD, 74.3 no-think (Table 3) | Capabilities |
| 15 Apr 2025 | Gemini 2.0 Flash model card | 63.5 | HiddenMath (Table, p.3, column Gemini 2.0 Flash (GA)) | Capabilities |
| 10 Apr 2025 | Gemini 2.0 Flash-Lite model card | 55.3 | HiddenMath (Table, p.3) | Capabilities |
| 25 Mar 2025 | Gemma 3 4B technical report | 43 | HiddenMath, IT, zero-shot (Table 6). Table 18 gives 42.0 | Capabilities |
| 25 Mar 2025 | Gemma 3 27B technical report | 60.3 | HiddenMath, IT, zero-shot (Table 6). Table 18 gives 56.0 | Capabilities |
| 25 Mar 2025 | Gemma 3 1B technical report | 15.8 | HiddenMath, IT, zero-shot (Table 6). Table 18 gives 15.0 | Capabilities |
| 25 Mar 2025 | Gemma 3 12B technical report | 54.5 | HiddenMath, IT, zero-shot (Table 6). Table 18 gives 51.0 | Capabilities |
Versions
More in Math
- IMO-AnswerBench In use 400 olympiad problems with short verifiable answers, robustified against memorization. Scored by answer match. GLM-5.2 91%
- AIME 2026 In use Models solve the 30 integer-answer problems from the 2026 AIME I and II math competitions. Scored by exact answer match. GPT-5.5 100%
- HMMT 2026 In use Models solve the 33 final-answer problems from the February 2026 Harvard-MIT Mathematics Tournament. Scored by exact answer match. GPT-5.5 98.5%
- MathArena (Apex, IMO, USAMO) In use Evaluates models on newly released math competitions after release. Final-answer or human/LLM-judged proof scores. Claude Opus 4.8 81.3%
- PolyMath In use Models solve 9,000 parallel math problems in 18 languages across four difficulty levels, from school math to frontier math. Final answers are checked.
Qwen3.7-Max 86.5% - USAMO 2026 In use Models write full proofs for the six problems of the 2026 USA Mathematical Olympiad. Judges grade each proof from 0 to 7 against rubrics. GPT-5.5 98.2%