Gemini 3 Pro Preview

Google·Operational·Released 2025-11-18textvisionaudiovideo
Context1,048.576k
Max output65.536k
Input / 1M$2 ↗
Output / 1M$12 ↗
Cached in-
Benchmarks25

Benchmarks

Cited public sources · rank is exact-benchmark, apples-to-apples

Chess Puzzles (Epoch AI run)#31/9231% ↗
SAGE (Vals AI)#31/7747.6% ↗
Vending-Bench 2 (Andon Labs)#24/595478.16 ↗
WeirdML (Håvard Tveit Ihle)#19/9269.9% ↗
BALROG (BALROG)#4/1958.1% ↗
GSO Opt@1 (GSO)#14/2717.7% ↗
AIME 2025#5/2495% ↗
ARC-AGI-2 (ARC Prize Verified)31.1% ↗
CharXiv Reasoning#3/481.4% ↗
GPQA Diamond#18/8391.9% ↗
MathArena Apex#2/223.4% ↗
MMMLU#1/991.8% ↗
MMMU-Pro#4/1281% ↗
SimpleBench (SimpleBench)#12/7376.4% ↗
Terminal-Bench 2.0 (Terminus-2 agent)54.2% ↗
Video-MMMU87.6% ↗
τ2-bench85.4% ↗

Gemini 3 Pro Preview FAQ

What is Gemini 3 Pro Preview?
Gemini 3 Pro Preview is a large language model from Google, released on November 18, 2025. It accepts text, image, audio and video input.
When was Gemini 3 Pro Preview released?
Google released Gemini 3 Pro Preview on November 18, 2025.
How much does Gemini 3 Pro Preview cost?
Gemini 3 Pro Preview costs $2 per 1M input tokens and $12 per 1M output tokens through the Google API. Prices are from Google's official pricing as of October 5, 2026.
What is Gemini 3 Pro Preview's context window?
Gemini 3 Pro Preview has a 1,049K-token context window and can generate up to 66K output tokens.
How does Gemini 3 Pro Preview score on benchmarks?
Gemini 3 Pro Preview has 25 cited benchmark scores, including Chess Puzzles (Epoch AI run) 31%, GPQA Diamond (Epoch AI run) 92.6% and OTIS Mock AIME 2024-2025 (Epoch AI run) 91.4%. Each score links to its source.

Access Gemini 3 Pro Preview and every other model through one endpoint with automatic failover: Respan gateway.