Skip to main content

Gemini 4 Carbon vs Gemini 4 Neon

SpecGemini 4 CarbonGemini 4 Neon
Respan Index76.1 (#3)79.2 (#2)
Blended price per 1M$4$6
Context window1M1M
Input / output per 1M$2 / $10$3 / $15
Max output-66K
Inputtext, image, audio, videotext, image, audio, video
Released2026-11-102026-11-18

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 4 CarbonGemini 4 Neon
Coding69.3 Gemini 4 Carbon78.0* Gemini 4 Neon
Agents83.5 Gemini 4 Carbon84.5 Gemini 4 Neon
Reasoning84.1 Gemini 4 Carbon87.8 Gemini 4 Neon
Knowledge & language77.8 Gemini 4 Carbon75.3 Gemini 4 Neon
Human preference60.6 Gemini 4 Carbon60.4 Gemini 4 Neon

* Estimated: no published scores in that category. How the Respan Index works

Benchmarks both report

Gemini 4 Neon leads on 8 of the 16 benchmarks both report. Higher is better. Each score links to where it was published.

BenchmarkGemini 4 CarbonGemini 4 Neon
LMArena Elo15271513
LMArena WebDev (LMArena)17691763
LMArena Agent (LMArena)0.1610.1841
APEX-Agents (Mercor)84.2%82.9%
Chess Puzzles (Epoch AI)78%83%
FrontierMath Tier 4 v2 (Epoch AI)95.1%97.6%
FrontierMath Tiers 1-3 v2 (Epoch AI)90.5%91.2%
Furniture Assembly (Epoch AI)77.5%89.2%
GPQA Diamond96.5%96.3%
Mystery Game Puzzles (Epoch AI)73%85%
OTIS Mock AIME 2024-2025 (Epoch AI)100%100%
SimpleQA Verified77.7%75.2%
Vending-Bench 2 (Andon Labs)16484.2116843.41
Agents' Last Exam v1.177.4%75.8%
GraphWalks 256k-1M BFS (F1)96.2%95.2%
SimpleBench (SimpleBench)92.6%94.4%

FAQ

Is Gemini 4 Carbon better than Gemini 4 Neon?

Gemini 4 Neon scores 79.2 on the Respan Index (#2), ahead of Gemini 4 Carbon at 76.1 (#3). Gemini 4 Neon leads on 8 of the 16 benchmarks both report.

Which is cheaper, Gemini 4 Carbon or Gemini 4 Neon?

Gemini 4 Carbon lists at $2 input and $10 output per 1M tokens; Gemini 4 Neon at $3 and $15. Gemini 4 Carbon is 1.5x cheaper per token at list price.

Which has the larger context window?

Both accept 1M tokens of context.

Compare other models

vs

Built for AI agents.
Break less.
Ship more.