Gemini 2.5 Pro vs Gemini 3 Flash preview

Gemini 2.5 Pro vs Gemini 3 Flash preview: Gemini 3 Flash preview is cheaper by 70% on average. Gemini 2.5 Pro from Google Vertex AI (1,048,576-token context, reasoning, tool calls) vs. Gemini 3 Flash preview from Google Vertex AI (1,048,576-token context, reasoning, tool calls). Use Agent Command Center to A/B both in shadow mode and pick the winner per workload.

Side-by-side cost

Live workload comparison

Same workload run through both models. The cheaper one is highlighted.

3,000
01,048,576
400
065,535
5,000
01,000,000
Google Vertex AI
$1,179/mo
Input $1.25/M · Output $10.00/M
Google Vertex AI
$411/mo
Input $0.500/M · Output $3.00/M
At this workload, Gemini 3 Flash preview is 65% cheaper than Gemini 2.5 Pro — a savings of $769/month ($9,223/year).
Production recipe — Agent Command Center
strategy: cost-optimized
primary:
  model: gemini-3-flash-preview
  provider: vertex-ai
fallback:
  model: gemini-2-5-pro
  provider: vertex-ai
shadow: { sample_rate: 0.05 }   # mirror 5% of traffic to compare quality live
Gemini 2.5 Pro Gemini 3 Flash preview
Input price $1.25/M $0.500/M
Output price $10.00/M $3.00/M
Context window 1,048,576 1,048,576
Max output 65,535 65,535
Function calling
Vision
Audio input
Reasoning
Prompt caching
Structured output
Pricing verified May 12, 2026 May 12, 2026
Cheaper option
~70% cheaper than Gemini 2.5 Pro
Larger context
1,048,576 tokens
More capabilities
6 of 6 capability flags advertised

Benchmark comparison

Side-by-side public benchmark scores. Greener bar = winner.

Chatbot Arena ELOgeneral
Gemini 2.5 Pro
1,448
Gemini 3 Flash preview
MATH-500math
Gemini 2.5 Pro
93.7%
Gemini 3 Flash preview
HumanEvalcode
Gemini 2.5 Pro
93.6%
Gemini 3 Flash preview
AIME 2025math
Gemini 2.5 Pro
86.7%
Gemini 3 Flash preview
MMLU-Proreasoning
Gemini 2.5 Pro
86.7%
Gemini 3 Flash preview
GPQA Diamondreasoning
Gemini 2.5 Pro
84.0%
Gemini 3 Flash preview
MMMUmultimodal
Gemini 2.5 Pro
79.6%
Gemini 3 Flash preview
BFCL v3agent
Gemini 2.5 Pro
76.0%
Gemini 3 Flash preview
Aider Polyglotcode
Gemini 2.5 Pro
73.3%
Gemini 3 Flash preview
LiveCodeBenchcode
Gemini 2.5 Pro
69.0%
Gemini 3 Flash preview
SWE-bench Verifiedagent
Gemini 2.5 Pro
63.8%
Gemini 3 Flash preview
Humanity's Last Examreasoning
Gemini 2.5 Pro
18.8%
Gemini 3 Flash preview