The Model Index
Up to four models

Gemini 4 Carbon vs Gemini 4 Argon

On the index (GPT-4 = 100), Gemini 4 Carbon scores 198 and Gemini 4 Argon 191. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.

Next to Gemini 4 Carbon
Presets
Gemini 4 Carbon
198182–215
WeightsClosed
ReleasedOct 2026
Price in / out$2.0 / $10
Context1.05M
Core tests taken4 of 9
Gemini 4 Argon
191176–207
WeightsClosed
ReleasedSep 2026
Price in / out$2.0 / $10
Context1.05M
Core tests taken6 of 9

Gemini 4 Carbon and Gemini 4 Argon are too close to call (7 points apart, within the uncertainty), ahead on 4 of 4 shared core tests, at a similar price.

Test by testthird-party results

Gemini 4 CarbonGemini 4 Argon
0%25%50%75%100%REASONING & KNOWLEDGEGPQA Diamond (Epoch AI)GPQA Diamond (Epoch AI)EBR-benchEBR-benchSimpleBenchSimpleBenchSimpleQA VerifiedSimpleQA VerifiedMATHSFrontierMath (tiers 1-3)FrontierMath (tiers 1-3)AIME-style maths (OTIS mock)AIME-style maths (OTIS mock)Chess puzzlesChess puzzlesFrontierMath (tier 4)FrontierMath (tier 4)CODINGTerminal-Bench 4.0Terminal-Bench 4.0Vibe Code Bench v1.1Vibe Code Bench v1.1FrontierSWEFrontierSWEAGENTSAPEX-AgentsAPEX-AgentsFurniture assemblyFurniture assemblyMystery game puzzlesMystery game puzzlesPROFESSIONAL WORKHLAB-Vals-1.1HLAB-Vals-1.1Legal Research BenchLegal Research BenchTax Agent BenchTax Agent Bench
Every result as a table
TestGemini 4 CarbonGemini 4 Argon
Reasoning & knowledge
GPQA Diamond (Epoch AI)96.5%96.4%
EBR-bench87.6%64.8%
SimpleBench92.6%91.7%
SimpleQA Verified77.7%76.0%
Maths
FrontierMath (tiers 1-3)90.5%86.0%
AIME-style maths (OTIS mock)100.0%100.0%
Chess puzzles78.0%59.0%
FrontierMath (tier 4)95.1%78.0%
Coding
Terminal-Bench 4.063.1%57.6%
Vibe Code Bench v1.189.1%91.9%
FrontierSWE61.2%55.0%
Agents
APEX-Agents84.2%82.4%
Furniture assembly77.5%72.5%
Mystery game puzzles73.0%64.0%
Professional work
HLAB-Vals-1.117.5%27.1%
Legal Research Bench55.8%54.8%
Tax Agent Bench51.8%49.2%

Tests at least two of these models have taken. Bold rows are the core tests.

This page as a chartDownload PNG