Skip to main content

Gemini 3 Flash Preview vs Gemini 4 Hydrogen

SpecGemini 3 Flash PreviewGemini 4 Hydrogen
Respan Index41.0 (#65)70.8 (#11)
Blended price per 1M$1.13$3
Context window1.0M1M
Input / output per 1M$0.5 / $3$1.5 / $7.5
Max output66K66K
Inputtext, vision, audiotext, vision, audio, video
Released2025-12-172026-11-12

Respan Index by category

Each category is scored 0 to 100, and how far ahead the winner is. Leads under a point are close to even.

CategoryGemini 3 Flash PreviewGemini 4 Hydrogen
Coding11.8 Gemini 3 Flash Preview62.5 Gemini 4 Hydrogen
Agents51.1 Gemini 3 Flash Preview78.7 Gemini 4 Hydrogen
Reasoning42.8 Gemini 3 Flash Preview77.9 Gemini 4 Hydrogen
Knowledge & language66.6 Gemini 3 Flash Preview71.0 Gemini 4 Hydrogen
Human preference52.5 Gemini 3 Flash Preview59.3 Gemini 4 Hydrogen

How the Respan Index works

Benchmarks both report

Gemini 4 Hydrogen leads on 11 of the 11 benchmarks both report. Higher is better. Each score links to where it was published.

BenchmarkGemini 3 Flash PreviewGemini 4 Hydrogen
LMArena Elo14731490
LMArena WebDev (LMArena)14401703
Chess Puzzles (Epoch AI)40%60%
FrontierMath Tier 4 v2 (Epoch AI)17.1%87.8%
FrontierMath Tiers 1-3 v2 (Epoch AI)51.2%88.1%
GPQA Diamond90.4%96.8%
Mystery Game Puzzles (Epoch AI)26%65%
OTIS Mock AIME 2024-2025 (Epoch AI)95.6%100%
SimpleQA Verified68.7%70.9%
Vending-Bench 2 (Andon Labs)3634.7214549.49
SimpleBench (SimpleBench)61.1%91.7%

FAQ

Is Gemini 3 Flash Preview better than Gemini 4 Hydrogen?

Gemini 4 Hydrogen scores 70.8 on the Respan Index (#11), ahead of Gemini 3 Flash Preview at 41.0 (#65). Gemini 4 Hydrogen leads on 11 of the 11 benchmarks both report.

Which is cheaper, Gemini 3 Flash Preview or Gemini 4 Hydrogen?

Gemini 3 Flash Preview lists at $0.5 input and $3 output per 1M tokens; Gemini 4 Hydrogen at $1.5 and $7.5. Gemini 3 Flash Preview is 2.7x cheaper per token at list price.

Which has the larger context window?

Gemini 3 Flash Preview accepts 1.0M tokens and Gemini 4 Hydrogen 1M tokens.

Compare other models

vs

Built for AI agents.
Break less.
Ship more.