Compare · ModelsLive · 2 picked · head to head

Gemini 3.1 Pro Preview vs GPT-5.2

Side by side · benchmarks, pricing, and signals you can act on.

Winner summary

Gemini 3.1 Pro Preview wins 19 of 30 shared benchmarks. Leads in agentic · reasoning · arena.

Category leads
agentic·Gemini 3.1 Pro Previewreasoning·Gemini 3.1 Pro Previewarena·Gemini 3.1 Pro Previewknowledge·Gemini 3.1 Pro Previewgeneral·Gemini 3.1 Pro Previewmath·GPT-5.2coding·GPT-5.2

Head to head · from the data · data as of 2026-11-17

  • Benchmarks · Gemini 3.1 Pro Preview leads on 19 of 30 shared benchmarks (APEX-Agents, ARC-AGI, ARC-AGI-2, Chatbot Arena Elo · Coding, 15 more), GPT-5.2 on 11 (Ebr Bench, FrontierMath-2025-02-28-Private, FrontierMath-Tier-4-2025-07-01-Private, FrontierMath-Tier-4-v2-Private, 7 more)
  • Price · Gemini 3.1 Pro Preview input $2.00/M vs GPT-5.2 $1.75/M, GPT-5.2 is 13% cheaper (OpenRouter list price, as of 2026-11-17)
  • Cheapest provider · Gemini 3.1 Pro Preview at Google and 1 other provider $1.00/M input (of 6 offers) as of 2026-11-17 · GPT-5.2 at OpenAI $0.88/M input (of 4 offers) as of 2026-11-17
  • Context · Gemini 3.1 Pro Preview 1.0M tokens vs GPT-5.2 400K tokens
  • Released · Gemini 3.1 Pro Preview Feb 19, 2026 vs GPT-5.2 Dec 10, 2025 · 2 months apart
  • Price history · Gemini 3.1 Pro Preview OpenRouter input list price had no move of 5% or more in the 90 days to 2026-11-17 · GPT-5.2 OpenRouter input list price had no move of 5% or more in the 90 days to 2026-11-17

Sources · scores from public leaderboards, list prices and provider offers from OpenRouter, price moves from the BenchGecko changelog (5% or more, held in the next daily check), Gecko tests run by BenchGecko.

Hype vs Reality
Gemini 3.1 Pro Preview
#143 by perf·#9 by attention
DESERVED
GPT-5.2
#168 by perf·#11 by attention
QUIET
Best value
1.2x better value than GPT-5.2
Gemini 3.1 Pro Preview
7.0 pts/$
$7.00/M
GPT-5.2
5.8 pts/$
$7.88/M
Vendor risk
Google DeepMind logo
Google DeepMind
$3.49T·Tier 1
Low risk
OpenAI logo
OpenAI
$840.0B·Tier 1
Medium risk
Head to head
Gemini 3.1 Pro PreviewGPT-5.2
APEX-Agents
Gemini 3.1 Pro Preview leads by +0.8
APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments.
Gemini 3.1 Pro Preview
35.1
GPT-5.2
34.3
ARC-AGI
Gemini 3.1 Pro Preview leads by +11.8
ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization.
Gemini 3.1 Pro Preview
98.0
GPT-5.2
86.2
ARC-AGI-2
Gemini 3.1 Pro Preview leads by +24.2
ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data.
Gemini 3.1 Pro Preview
77.1
GPT-5.2
52.9
Chatbot Arena Elo · Coding
Gemini 3.1 Pro Preview leads by +30.7
Gemini 3.1 Pro Preview
1446.2
GPT-5.2
1415.5
Chatbot Arena Elo · Overall
Gemini 3.1 Pro Preview leads by +51.4
Gemini 3.1 Pro Preview
1487.0
GPT-5.2
1435.6
Chess Puzzles
Gemini 3.1 Pro Preview leads by +3.1
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
Gemini 3.1 Pro Preview
48.4
GPT-5.2
45.3
Cl Bench
Gemini 3.1 Pro Preview leads by +2.6
Gemini 3.1 Pro Preview
20.8
GPT-5.2
18.2
DeepResearch Bench
Gemini 3.1 Pro Preview leads by +6.7
DeepResearch Bench · evaluates AI on complex multi-step research tasks requiring information gathering, synthesis, and producing comprehensive analyses.
Gemini 3.1 Pro Preview
47.8
GPT-5.2
41.1
Dtbench
Gemini 3.1 Pro Preview leads by +10.2
Gemini 3.1 Pro Preview
95.1
GPT-5.2
84.9
Ebr Bench
GPT-5.2 leads by +8.7
Gemini 3.1 Pro Preview
14.3
GPT-5.2
23.0
FrontierMath-2025-02-28-Private
GPT-5.2 leads by +3.8
FrontierMath (Feb 2025) · original research-level math problems created by mathematicians, testing capabilities at the boundary of current AI mathematical reasoning.
Gemini 3.1 Pro Preview
36.9
GPT-5.2
40.7
FrontierMath-Tier-4-2025-07-01-Private
GPT-5.2 leads by +2.1
FrontierMath Tier 4 (Jul 2025) · the most challenging tier of frontier mathematics, containing problems that push the absolute limits of AI mathematical reasoning.
Gemini 3.1 Pro Preview
16.7
GPT-5.2
18.8
FrontierMath-Tier-4-v2-Private
GPT-5.2 leads by +4.9
Gemini 3.1 Pro Preview
26.8
GPT-5.2
31.7
FrontierMath-Tiers-1-3-v2-Private
GPT-5.2 leads by +8.2
Gemini 3.1 Pro Preview
62.8
GPT-5.2
71.0
Furniture Assembly
GPT-5.2 leads by +11.9
Gemini 3.1 Pro Preview
0.0
GPT-5.2
11.9
GPQA diamond
Gemini 3.1 Pro Preview leads by +4.2
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
Gemini 3.1 Pro Preview
96.5
GPT-5.2
92.2
GSO-Bench
GPT-5.2 leads by +4.8
GSO-Bench · evaluates AI models on real-world open-source software engineering tasks, testing the ability to understand and resolve actual GitHub issues.
Gemini 3.1 Pro Preview
22.6
GPT-5.2
27.4
HLE
Gemini 3.1 Pro Preview leads by +19.6
HLE (Humanity's Last Exam) · a reasoning benchmark designed to be the hardest public evaluation of AI. Questions span mathematics, physics, philosophy, and logic · curated to be at or beyond the frontier of human expert capability. Tested with and without tool augmentation.
Gemini 3.1 Pro Preview
43.7
GPT-5.2
24.2
Lmca
Gemini 3.1 Pro Preview leads by +11.7
Gemini 3.1 Pro Preview
63.3
GPT-5.2
51.7
Metr Time Horizons
Gemini 3.1 Pro Preview leads by +1.8
Gemini 3.1 Pro Preview
77.0
GPT-5.2
75.3
Mirrorcode
GPT-5.2 leads by +23.3
Gemini 3.1 Pro Preview
8.9
GPT-5.2
32.2
Mystery Game Puzzles
Gemini 3.1 Pro Preview leads by +12.1
Gemini 3.1 Pro Preview
27.3
GPT-5.2
15.2
OTIS Mock AIME 2024-2025
GPT-5.2 leads by +0.5
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
Gemini 3.1 Pro Preview
95.6
GPT-5.2
96.1
PostTrainBench
GPT-5.2 leads by +3.3
Gemini 3.1 Pro Preview
18.1
GPT-5.2
21.4
Proofbench
Gemini 3.1 Pro Preview leads by +11.0
Gemini 3.1 Pro Preview
26.0
GPT-5.2
15.0
SimpleBench
Gemini 3.1 Pro Preview leads by +40.6
SimpleBench · tests fundamental reasoning capabilities with straightforward problems designed to expose gaps in basic logical and spatial thinking.
Gemini 3.1 Pro Preview
75.5
GPT-5.2
35.0
SimpleQA Verified
Gemini 3.1 Pro Preview leads by +34.6
SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information.
Gemini 3.1 Pro Preview
73.5
GPT-5.2
38.9
SWE-Bench verified
Gemini 3.1 Pro Preview leads by +1.9
SWE-bench Verified · 500 human-validated tasks from 12 real Python repositories (Django, Flask, scikit-learn, sympy, and others). Each task requires the model to produce a git patch that resolves a real GitHub issue and passes the test suite. The verified subset eliminates ambiguous tasks from the original SWE-bench. The benchmark remains the most-cited evaluation for code-generation capability.
Gemini 3.1 Pro Preview
75.6
GPT-5.2
73.8
Terminal Bench
Gemini 3.1 Pro Preview leads by +15.3
Terminal-Bench 2.0 · evaluates AI agents on real terminal-based coding tasks · writing scripts, debugging, running tests, and managing projects entirely through command-line interaction. Tests both code quality and terminal fluency.
Gemini 3.1 Pro Preview
80.2
GPT-5.2
64.9
WeirdML
GPT-5.2 leads by +0.1
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
Gemini 3.1 Pro Preview
72.1
GPT-5.2
72.2
Full benchmark table
BenchmarkGemini 3.1 Pro PreviewGPT-5.2
APEX-Agents
APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments.
35.134.3
ARC-AGI
ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization.
98.086.2
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.
77.152.9
Chatbot Arena Elo · Coding
1446.21415.5
Chatbot Arena Elo · Overall
1487.01435.6
Chess Puzzles
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
48.445.3
Cl Bench
20.818.2
DeepResearch Bench
DeepResearch Bench · evaluates AI on complex multi-step research tasks requiring information gathering, synthesis, and producing comprehensive analyses.
47.841.1
Dtbench
95.184.9
Ebr Bench
14.323.0
FrontierMath-2025-02-28-Private
FrontierMath (Feb 2025) · original research-level math problems created by mathematicians, testing capabilities at the boundary of current AI mathematical reasoning.
36.940.7
FrontierMath-Tier-4-2025-07-01-Private
FrontierMath Tier 4 (Jul 2025) · the most challenging tier of frontier mathematics, containing problems that push the absolute limits of AI mathematical reasoning.
16.718.8
FrontierMath-Tier-4-v2-Private
26.831.7
FrontierMath-Tiers-1-3-v2-Private
62.871.0
Furniture Assembly
0.011.9
GPQA diamond
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
96.592.2
GSO-Bench
GSO-Bench · evaluates AI models on real-world open-source software engineering tasks, testing the ability to understand and resolve actual GitHub issues.
22.627.4
HLE
HLE (Humanity's Last Exam) · a reasoning benchmark designed to be the hardest public evaluation of AI. Questions span mathematics, physics, philosophy, and logic · curated to be at or beyond the frontier of human expert capability. Tested with and without tool augmentation.
43.724.2
Lmca
63.351.7
Metr Time Horizons
77.075.3
Mirrorcode
8.932.2
Mystery Game Puzzles
27.315.2
OTIS Mock AIME 2024-2025
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
95.696.1
PostTrainBench
18.121.4
Proofbench
26.015.0
SimpleBench
SimpleBench · tests fundamental reasoning capabilities with straightforward problems designed to expose gaps in basic logical and spatial thinking.
75.535.0
SimpleQA Verified
SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information.
73.538.9
SWE-Bench verified
SWE-bench Verified · 500 human-validated tasks from 12 real Python repositories (Django, Flask, scikit-learn, sympy, and others). Each task requires the model to produce a git patch that resolves a real GitHub issue and passes the test suite. The verified subset eliminates ambiguous tasks from the original SWE-bench. The benchmark remains the most-cited evaluation for code-generation capability.
75.673.8
Terminal Bench
Terminal-Bench 2.0 · evaluates AI agents on real terminal-based coding tasks · writing scripts, debugging, running tests, and managing projects entirely through command-line interaction. Tests both code quality and terminal fluency.
80.264.9
WeirdML
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
72.172.2
Pricing · per 1M tokens · projected $/mo at 10M tokens
ModelInputOutputContextProjected $/mo
Google DeepMind logoGemini 3.1 Pro Preview$2.00$12.001.0M tokens (~524 books)$45.00
OpenAI logoGPT-5.2$1.75$14.00400K tokens (~200 books)$48.13
Gemini 3.1 Pro Preview leads on 19 of 30 shared benchmarks (APEX-Agents, ARC-AGI, ARC-AGI-2, Chatbot Arena Elo · Coding, 15 more), GPT-5.2 on 11 (Ebr Bench, FrontierMath-2025-02-28-Private, FrontierMath-Tier-4-2025-07-01-Private, FrontierMath-Tier-4-v2-Private, 7 more)