Compare · ModelsLive · 2 picked · head to head
Gemini 2.5 Pro vs Gemini 3.6 Flash
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Gemini 3.6 Flash wins on 12/12 benchmarks
Gemini 3.6 Flash wins 12 of 12 shared benchmarks. Leads in agentic · reasoning · knowledge.
Category leads
agentic·Gemini 3.6 Flashreasoning·Gemini 3.6 Flashknowledge·Gemini 3.6 Flashgeneral·Gemini 3.6 Flashmath·Gemini 3.6 Flashcoding·Gemini 3.6 Flash
Hype vs Reality
Attention vs performance
Gemini 2.5 Pro
#125 by perf·no signal
Gemini 3.6 Flash
#98 by perf·#10 by attention
Best value
Gemini 3.6 Flash
2.7x better value than Gemini 2.5 Pro
Gemini 2.5 Pro
9.0 pts/$
$5.63/M
Gemini 3.6 Flash
24.0 pts/$
$2.25/M
Vendor risk
Who is behind the model
Google DeepMind
$3.53T·Tier 1
Google DeepMind
$3.53T·Tier 1
Head to head
12 benchmarks · 2 models
Gemini 2.5 ProGemini 3.6 Flash
APEX-Agents
Gemini 3.6 Flash leads by +40.3
APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments.
Gemini 2.5 Pro
6.6
Gemini 3.6 Flash
46.9
ARC-AGI
Gemini 3.6 Flash leads by +50.2
ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization.
Gemini 2.5 Pro
41.0
Gemini 3.6 Flash
91.2
ARC-AGI-2
Gemini 3.6 Flash leads by +55.6
ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data.
Gemini 2.5 Pro
4.9
Gemini 3.6 Flash
60.4
Chess Puzzles
Gemini 3.6 Flash leads by +23.1
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
Gemini 2.5 Pro
15.8
Gemini 3.6 Flash
39.0
Dtbench
Gemini 3.6 Flash leads by +21.8
Gemini 2.5 Pro
70.7
Gemini 3.6 Flash
92.5
FrontierMath-Tier-4-v2-Private
Gemini 3.6 Flash leads by +21.9
Gemini 2.5 Pro
0.0
Gemini 3.6 Flash
21.9
FrontierMath-Tiers-1-3-v2-Private
Gemini 3.6 Flash leads by +36.2
Gemini 2.5 Pro
25.9
Gemini 3.6 Flash
62.0
GPQA diamond
Gemini 3.6 Flash leads by +12.3
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
Gemini 2.5 Pro
83.7
Gemini 3.6 Flash
96.0
Lmca
Gemini 3.6 Flash leads by +12.0
Gemini 2.5 Pro
40.9
Gemini 3.6 Flash
52.9
OTIS Mock AIME 2024-2025
Gemini 3.6 Flash leads by +9.5
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
Gemini 2.5 Pro
84.7
Gemini 3.6 Flash
94.2
SimpleQA Verified
Gemini 3.6 Flash leads by +10.2
SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information.
Gemini 2.5 Pro
56.0
Gemini 3.6 Flash
66.2
WeirdML
Gemini 3.6 Flash leads by +2.1
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
Gemini 2.5 Pro
54.0
Gemini 3.6 Flash
56.1
Full benchmark table
| Benchmark | Gemini 2.5 Pro | Gemini 3.6 Flash |
|---|---|---|
APEX-Agents APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments. | 6.6 | 46.9 |
ARC-AGI ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization. | 41.0 | 91.2 |
ARC-AGI-2 ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data. | 4.9 | 60.4 |
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | 15.8 | 39.0 |
Dtbench | 70.7 | 92.5 |
FrontierMath-Tier-4-v2-Private | 0.0 | 21.9 |
FrontierMath-Tiers-1-3-v2-Private | 25.9 | 62.0 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 83.7 | 96.0 |
Lmca | 40.9 | 52.9 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 84.7 | 94.2 |
SimpleQA Verified SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information. | 56.0 | 66.2 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 54.0 | 56.1 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $1.25 | $10.00 | 1.0M tokens (~524 books) | $34.38 | |
| $0.75 | $3.75 | 1.0M tokens (~524 books) | $15.00 |