Gemini 3.5 Flash vs Gemini 3.6 Flash
Different efforts: medium · high · Compare all at high
Humanity's Last Exam Humanity's Last Exam: Very hard expert questions across many fields, written to stump today’s models
Gemini 3.5 Flash41.3
Gemini 3.6 Flash40.8
Arena Text · overall Arena Text · overall: Human pairwise preferences in the stated text vote pool
Gemini 3.5 Flash1,476
Gemini 3.6 Flash1,483
Arena Text · Chinese Arena Text · chinese: Human pairwise preferences in the stated text vote pool
Gemini 3.5 Flash1,526.4
Gemini 3.6 Flash1,530.4
Arena Text · English Arena Text · english: Human pairwise preferences in the stated text vote pool
Gemini 3.5 Flash1,476.2
Gemini 3.6 Flash1,481.7
Arena Text · French Arena Text · french: Human pairwise preferences in the stated text vote pool
Gemini 3.5 Flash1,475.7
Gemini 3.6 Flash1,502.5
Arena Text · German Arena Text · german: Human pairwise preferences in the stated text vote pool
Gemini 3.5 Flash1,493.3
Gemini 3.6 Flash1,481
Arena Text · instruction following Arena Text · instruction following: Human pairwise preferences in the stated text vote pool
Gemini 3.5 Flash1,464.2
Gemini 3.6 Flash1,466.7
Arena Text · Japanese Arena Text · japanese: Human pairwise preferences in the stated text vote pool
Gemini 3.5 Flash1,461.6
Gemini 3.6 Flash1,476.3
Arena Text · Korean Arena Text · korean: Human pairwise preferences in the stated text vote pool
Gemini 3.5 Flash1,449.2
Gemini 3.6 Flash1,430
Arena Text · non-English Arena Text · non english: Human pairwise preferences in the stated text vote pool
Gemini 3.5 Flash1,471
Gemini 3.6 Flash1,469.5
Arena Text · Polish Arena Text · polish: Human pairwise preferences in the stated text vote pool
Gemini 3.5 Flash1,499.7
Gemini 3.6 Flash1,474.8
Arena Text · Russian Arena Text · russian: Human pairwise preferences in the stated text vote pool
Gemini 3.5 Flash1,491.2
Gemini 3.6 Flash1,488
Arena Text · Spanish Arena Text · spanish: Human pairwise preferences in the stated text vote pool
Gemini 3.5 Flash1,475.1
Gemini 3.6 Flash1,475.9
AA Omniscience Index AA Omniscience Index: Scores factual knowledge while penalizing incorrect answers to encourage admitting uncertainty
Gemini 3.5 Flash20.8
Gemini 3.6 Flash22.1
SimpleQA Verified v1 SimpleQA Verified v1: Accuracy on short factual questions with verified answers, answered without web access
Gemini 3.5 Flash70.4
Gemini 3.6 Flash66.2
CritPt (physics) CritPt (physics): Research-level physics problems that need many careful reasoning steps
Gemini 3.5 Flash10.9
Gemini 3.6 Flash10.6
AA Intelligence Index 4.3.2 AA Intelligence Index 4.3.2: Artificial Analysis combines tests of knowledge, reasoning, coding and tool use into one score
Gemini 3.5 Flash33.6
Gemini 3.6 Flash34
AA Omniscience · Accuracy AA Omniscience · Accuracy: Measures correct answers on the AA Omniscience factual knowledge test
Gemini 3.5 Flash51.1
Gemini 3.6 Flash50
AA Omniscience · Hallucination Rate AA Omniscience · Hallucination Rate: Measures how often attempted answers are incorrect on the AA Omniscience factual knowledge test · lower is better
Gemini 3.5 Flash61.8
Gemini 3.6 Flash55.6
AA-LCR (long context) 1.1 AA-LCR (long context) 1.1: Questions that need reading and connecting facts across very long documents (around 100K tokens)
Gemini 3.5 Flash74.3
Gemini 3.6 Flash80
GDP.pdf · AA all-pass GDP.pdf · AA all-pass: Models reason over professional PDF documents; an attempt passes only when every atomic criterion passes
Gemini 3.5 Flash · high19.8
Gemini 3.6 Flash17.4
MLCR-AA MLCR-AA: Models synthesize long medical records; the headline credits concise answers judged both accurate and complete
Gemini 3.5 Flash · high18.3
Gemini 3.6 Flash14.4
ARC-AGI-1 · public evaluation ARC-AGI-1 · public-eval: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit
Gemini 3.5 Flash · high96
Gemini 3.6 Flash96
ARC-AGI-1 · semi-private ARC-AGI-1 · semi-private: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit
Gemini 3.5 Flash · high92.5
Gemini 3.6 Flash91.2
ARC-AGI-2 · public evaluation ARC-AGI-2 · public-eval: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit
Gemini 3.5 Flash · high72.1
Gemini 3.6 Flash65.3
ARC-AGI-2 · semi-private ARC-AGI-2 · semi-private: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit
Gemini 3.5 Flash · high72.1
Gemini 3.6 Flash60.4
Chess Puzzles Chess Puzzles: Models choose the best next move from a novel chess position
Gemini 3.5 Flash · high45
Gemini 3.6 Flash40
FrontierMath Tier 4 (v2) FrontierMath Tier 4 (v2): Models solve expert-written advanced mathematics problems on the named tier and edition
Gemini 3.5 Flash · high26.8
Gemini 3.6 Flash22
FrontierMath Tiers 1–3 (v2) FrontierMath Tiers 1–3 (v2): Models solve expert-written advanced mathematics problems on the named tier and edition
Gemini 3.5 Flash · high62.8
Gemini 3.6 Flash58.9
Mystery Game Puzzles Mystery Game Puzzles: Models choose the best move in puzzle positions from an undisclosed game
Gemini 3.5 Flash · high32
Gemini 3.6 Flash30
MathArena · Apex MathArena · apex · Apex: Answer accuracy on the named mathematics competition under the recorded effort
Gemini 3.5 Flash32.3
Gemini 3.6 Flash26
MathArena · Apex Shortlist MathArena · apex · Apex Shortlist: Answer accuracy on the named mathematics competition under the recorded effort
Gemini 3.5 Flash82.5
Gemini 3.6 Flash71.3
MathArena · ArXivMath · April 2026 MathArena · arxiv · 04/2026: Answer accuracy on the named mathematics competition under the recorded effort
Gemini 3.5 Flash52.5
Gemini 3.6 Flash52.5
MathArena · ArXivMath · June 2026 MathArena · arxiv · 06/2026: Answer accuracy on the named mathematics competition under the recorded effort
Gemini 3.5 Flash52.1
Gemini 3.6 Flash56.9
MathArena · ArXivMath · May 2026 MathArena · arxiv · 05/2026: Answer accuracy on the named mathematics competition under the recorded effort
Gemini 3.5 Flash44.2
Gemini 3.6 Flash54.2
MathArena · BrokenArXiv · April 2026 MathArena · arxiv_false · 04/2026: Answer accuracy on the named mathematics competition under the recorded effort
Gemini 3.5 Flash21.3
Gemini 3.6 Flash16
MathArena · BrokenArXiv · June 2026 MathArena · arxiv_false · 06/2026: Answer accuracy on the named mathematics competition under the recorded effort
Gemini 3.5 Flash12.4
Gemini 3.6 Flash12
MathArena · BrokenArXiv · May 2026 MathArena · arxiv_false · 05/2026: Answer accuracy on the named mathematics competition under the recorded effort
Gemini 3.5 Flash11
Gemini 3.6 Flash4
MathArena · Kangaroo 2025 1-2 MathArena · kangaroo · Kangaroo 2025 1-2: Answer accuracy on the named mathematics competition under the recorded effort
Gemini 3.5 Flash89.6
Gemini 3.6 Flash83.3
MathArena · Kangaroo 2025 11-12 MathArena · kangaroo · Kangaroo 2025 11-12: Answer accuracy on the named mathematics competition under the recorded effort
Gemini 3.5 Flash93.3
Gemini 3.6 Flash96.7
MathArena · Kangaroo 2025 3-4 MathArena · kangaroo · Kangaroo 2025 3-4: Answer accuracy on the named mathematics competition under the recorded effort
Gemini 3.5 Flash72.9
Gemini 3.6 Flash68.8
MathArena · Kangaroo 2025 5-6 MathArena · kangaroo · Kangaroo 2025 5-6: Answer accuracy on the named mathematics competition under the recorded effort
Gemini 3.5 Flash90
Gemini 3.6 Flash88.3
MathArena · Kangaroo 2025 7-8 MathArena · kangaroo · Kangaroo 2025 7-8: Answer accuracy on the named mathematics competition under the recorded effort
Gemini 3.5 Flash93.3
Gemini 3.6 Flash90
MathArena · Kangaroo 2025 9-10 MathArena · kangaroo · Kangaroo 2025 9-10: Answer accuracy on the named mathematics competition under the recorded effort
Gemini 3.5 Flash100
Gemini 3.6 Flash100
Vals LegalBench Vals LegalBench: Legal reasoning and legal-language understanding on LegalBench
Gemini 3.5 Flash83.6
Gemini 3.6 Flash86.7
Vals MedCode Vals MedCode: Medical coding accuracy on patient encounters
Gemini 3.5 Flash55.8
Gemini 3.6 Flash53.2
Vals MedScribe Vals MedScribe: Accuracy of clinical notes produced from consultation transcripts
Gemini 3.5 Flash76.6
Gemini 3.6 Flash79.7
Vals Legal Research Bench Vals Legal Research Bench: Grounded legal research; a task passes only if all criteria pass
Gemini 3.5 Flash30.8
Gemini 3.6 Flash25
IFBench (instructions) IFBench (instructions): Checks whether the model follows precise formatting and content instructions
Gemini 3.5 Flash74.6
Gemini 3.6 Flash—
FrontierMath FrontierMath: Models solve expert-written advanced mathematics problems on the named tier and edition
Gemini 3.5 Flash · high39
Gemini 3.6 Flash—
FrontierMath Tier 4 FrontierMath Tier 4: Models solve expert-written advanced mathematics problems on the named tier and edition
Gemini 3.5 Flash · high14.6
Gemini 3.6 Flash—
MathArena · ArXivMath · February 2026 MathArena · arxiv · 02/2026: Answer accuracy on the named mathematics competition under the recorded effort
Gemini 3.5 Flash50.8
Gemini 3.6 Flash—
MathArena · ArXivMath · March 2026 MathArena · arxiv · 03/2026: Answer accuracy on the named mathematics competition under the recorded effort
Gemini 3.5 Flash55.8
Gemini 3.6 Flash—
MathArena · BrokenArXiv · February 2026 MathArena · arxiv_false · 02/2026: Answer accuracy on the named mathematics competition under the recorded effort
Gemini 3.5 Flash6.5
Gemini 3.6 Flash—
MathArena · BrokenArXiv · March 2026 MathArena · arxiv_false · 03/2026: Answer accuracy on the named mathematics competition under the recorded effort
Gemini 3.5 Flash16.1
Gemini 3.6 Flash—
MathArena · Project Euler MathArena · euler · Project Euler: Answer accuracy on the named mathematics competition under the recorded effort
Gemini 3.5 Flash82
Gemini 3.6 Flash—
| Benchmark | Gemini 3.5 Flash | Gemini 3.6 Flash |
|---|---|---|
| Humanity's Last Exam | 41.3 | 40.8 |
| Arena Text · overall | 1,476 | 1,483 |
| Arena Text · Chinese | 1,526.4 | 1,530.4 |
| Arena Text · English | 1,476.2 | 1,481.7 |
| Arena Text · French | 1,475.7 | 1,502.5 |
| Arena Text · German | 1,493.3 | 1,481 |
| Arena Text · instruction following | 1,464.2 | 1,466.7 |
| Arena Text · Japanese | 1,461.6 | 1,476.3 |
| Arena Text · Korean | 1,449.2 | 1,430 |
| Arena Text · non-English | 1,471 | 1,469.5 |
| Arena Text · Polish | 1,499.7 | 1,474.8 |
| Arena Text · Russian | 1,491.2 | 1,488 |
| Arena Text · Spanish | 1,475.1 | 1,475.9 |
| AA Omniscience Index | 20.8 | 22.1 |
| SimpleQA Verified v1 | 70.4 | 66.2 |
| CritPt (physics) | 10.9 | 10.6 |
| AA Intelligence Index 4.3.2 | 33.6 | 34 |
| AA Omniscience · Accuracy | 51.1 | 50 |
| AA Omniscience · Hallucination Rate | 61.8 | 55.6 |
| AA-LCR (long context) 1.1 | 74.3 | 80 |
| GDP.pdf · AA all-pass | 19.8 · high | 17.4 |
| MLCR-AA | 18.3 · high | 14.4 |
| ARC-AGI-1 · public evaluation | 96 · high | 96 |
| ARC-AGI-1 · semi-private | 92.5 · high | 91.2 |
| ARC-AGI-2 · public evaluation | 72.1 · high | 65.3 |
| ARC-AGI-2 · semi-private | 72.1 · high | 60.4 |
| Chess Puzzles | 45 · high | 40 |
| FrontierMath Tier 4 (v2) | 26.8 · high | 22 |
| FrontierMath Tiers 1–3 (v2) | 62.8 · high | 58.9 |
| Mystery Game Puzzles | 32 · high | 30 |
| MathArena · Apex | 32.3 | 26 |
| MathArena · Apex Shortlist | 82.5 | 71.3 |
| MathArena · ArXivMath · April 2026 | 52.5 | 52.5 |
| MathArena · ArXivMath · June 2026 | 52.1 | 56.9 |
| MathArena · ArXivMath · May 2026 | 44.2 | 54.2 |
| MathArena · BrokenArXiv · April 2026 | 21.3 | 16 |
| MathArena · BrokenArXiv · June 2026 | 12.4 | 12 |
| MathArena · BrokenArXiv · May 2026 | 11 | 4 |
| MathArena · Kangaroo 2025 1-2 | 89.6 | 83.3 |
| MathArena · Kangaroo 2025 11-12 | 93.3 | 96.7 |
| MathArena · Kangaroo 2025 3-4 | 72.9 | 68.8 |
| MathArena · Kangaroo 2025 5-6 | 90 | 88.3 |
| MathArena · Kangaroo 2025 7-8 | 93.3 | 90 |
| MathArena · Kangaroo 2025 9-10 | 100 | 100 |
| Vals LegalBench | 83.6 | 86.7 |
| Vals MedCode | 55.8 | 53.2 |
| Vals MedScribe | 76.6 | 79.7 |
| Vals Legal Research Bench | 30.8 | 25 |
| IFBench (instructions) | 74.6 | — |
| FrontierMath | 39 · high | — |
| FrontierMath Tier 4 | 14.6 · high | — |
| MathArena · ArXivMath · February 2026 | 50.8 | — |
| MathArena · ArXivMath · March 2026 | 55.8 | — |
| MathArena · BrokenArXiv · February 2026 | 6.5 | — |
| MathArena · BrokenArXiv · March 2026 | 16.1 | — |
| MathArena · Project Euler | 82 | — |
Specs
| Spec | Gemini 3.5 Flash | Gemini 3.6 Flash |
|---|---|---|
| Input, per 1M | $1.50 | !$0.75 |
| Output, per 1M | $9 | !$3.75 |
| Cost per task? | — | $1.60 |
| Speed, tok/s | 199 | 209 |
| Context | 1M | 1M |
| Weights | Closed | Closed |
| Released | May 2026 | Jul 2026 |
| See images | ✓ | ✓ |
| Read PDFs | Partial | ✓ |
| Hear audio | ✓ | ✓ |
| Watch video | ✓ | ✓ |
| Call tools | ✓ | ✓ |
| Force a tool call | Partial | Partial |
| Return strict JSON | ✓ | ✓ |
| Search the web | ? | ✓ |
| Run code | ? | ✓ |
| Use a computer | ? | Partial |
What breaks when you switch
Gemini 3.5 Flash → Gemini 3.6 Flash
Nothing known.
Gemini 3.6 Flash → Gemini 3.5 Flash
Nothing known.