Skip to content

Gemini 4 Neon Benchmarks: Did Google Really Beat Claude?

Google has officially announced Gemini 4 Neon, its latest model, which competes with Claude Opus 5.5 and the recently announced GPT-6.2 Sol. On the benchmarks Google shared, Neon beats both Claude Opus 5.5 and GPT-6.2 Sol on most tests, some by a huge margin. In this post, we compare the new Neon model with the other two across plenty of AI benchmarks.

All benchmarks

BenchmarkGemini 4 NeonClaude Opus 5.5GPT-6.2 Sol
AA Index v5.0.1615957
Vals Index v2.270.9%70.1%66.1%
CursorBench 4.165.4%58.1%-
Terminal-Bench 4.165.2%65.8%61.5%
FrontierCode 2.046.2%47.3%43.8%
SWE-sweep (provider harness)39.0%10.1%8.3%
SWE-sweep (Mini-SWE harness)12.8%3.6%11.6%
AutomationBench56.3%42.5%41.3%
Agents' Last Exam v1.175.8%71.6%73.0%
RiemannBench84.0%80.0%80.0%
Terminal-Bench Science 0.170.5%63.3%68.0%
HLE-Diamond71.3%67.9%64.0%
MKA (Climb WSR)56.5%53.0%52.7%
Chartography73.5%66.3%75.1%
LVBench89.0%83.7%87.6%
GraphWalks (256K to 1M)95.2%66.8%80.1%
CWE-Bench v270.7%67.1%66.0%
SEC-bench Pro90.5%-88.3%
WebExploitBench v1.251.3%37.1%-

A dash means no score has been reported for that model.

Winner by category

CategoryWinner
Overall intelligence: AA IndexGemini 4 Neon
Professional work: Vals IndexGemini 4 Neon
CursorBenchGemini 4 Neon
Terminal-Bench 4.1Claude Opus 5.5
FrontierCode 2.0Claude Opus 5.5
SWE-sweepGemini 4 Neon
AutomationGemini 4 Neon
Professional agentsGemini 4 Neon
Mathematics: RiemannBenchGemini 4 Neon
Scientific agent workGemini 4 Neon
Hard reasoning: HLE-DiamondGemini 4 Neon
Multimodal agent workGemini 4 Neon
Chart reasoningGPT-6.2 Sol
Long video understandingGemini 4 Neon
Long context reasoningGemini 4 Neon
Vulnerability patchingGemini 4 Neon
Vulnerability discoveryGemini 4 Neon
Web exploitationGemini 4 Neon

Leaderboard positions

BenchmarkGemini 4 NeonRanking
Artificial Analysis Intelligence Index v5.0.161#3 overall
Artificial Analysis Coding Agent Index58.6#1
Vals Index v2.270.9%#2
Agent Arena Index1336#2
AutomationBench56.3%#1
Vending-Bench 2$16,843#1
Terminal-Bench Science 0.170.5%Best in class
RiemannBench84.0%Best in class
HLE-Diamond71.3%Best in class

Coding benchmarks

BenchmarkGemini 4 NeonClaude Opus 5.5GPT-6.2 Sol
CursorBench 4.165.4%58.1%-
Terminal-Bench 4.165.2%65.8%61.5%
FrontierCode 2.046.2%47.3%43.8%
SWE-sweep (provider harness)39.0%10.1%8.3%
SWE-sweep (Mini-SWE harness)12.8%3.6%11.6%

Science and reasoning

BenchmarkGemini 4 NeonClaude Opus 5.5GPT-6.2 Sol
Terminal-Bench Science 0.170.5%63.3%68.0%
RiemannBench84.0%80.0%80.0%
HLE-Diamond71.3%67.9%64.0%
Agents' Last Exam v1.175.8%71.6%73.0%

Multimodal and long context

BenchmarkGemini 4 NeonClaude Opus 5.5GPT-6.2 Sol
GraphWalks (256K to 1M)95.2%66.8%80.1%
LVBench89.0%83.7%87.6%
MKA (Climb WSR)56.5%53.0%52.7%
Chartography73.5%66.3%75.1%

Cybersecurity

BenchmarkGemini 4 NeonClaude Opus 5.5GPT-6.2 Sol
SEC-bench Pro90.5%-88.3%
CWE-Bench v270.7%67.1%66.0%
WebExploitBench v1.251.3%37.1%-
AutoBaxBench (hard tier)76%--

API pricing

ModelInput, per 1MCached input, per 1MOutput, per 1M
Gemini 4 Neon$3.00$0.15$15.00
GPT-6.2 Sol$2.00$0.10$10.00
Claude Opus 5.5$4.00$0.20$20.00

Thinking levels

SettingGemini 4 Neon
LowYes
MediumYes
HighYes, default
xhighYes, highest

Specifications

SpecificationGemini 4 Neon
Model IDgemini-4-neon
Model familyGemini 4
Base modelAdapted from Gemini 4 Carbon
Context window1M tokens, input and output combined
Default maximum output64K tokens
Background long decodeUp to 1M output tokens
Context in AntigravityUp to 512K tokens
Knowledge cutoffMarch 2026
Text inputYes
Image inputYes
Audio inputYes
Video inputYes
OutputText
ReasoningYes
Default thinkingHigh
Long-running agentsYes
Coding agentsYes
Multimodal knowledge workYes
Cybersecurity workflowsYes
Release dateNovember 18, 2026

Availability

PlatformGemini 4 Neon
Gemini APIPaid tier
Google AI StudioYes
Gemini appGoogle AI Pro and Ultra
Gemini for businessYes
Google AntigravityPaid Google AI Pro and Ultra
Orchestrator in AntigravityYes
Android StudioYes
Free Gemini app usersNo direct model selection
Google AI PlusNot included in Antigravity
Fairwind Program (cyber defenders)Rolling out to eligible defenders

Head-to-head summary

CategoryGemini 4 NeonClaude Opus 5.5GPT-6.2 SolWinner
Overall intelligence615957Neon
Knowledge work70.9%70.1%66.1%Neon
Cursor coding65.4%58.1%-Neon
Terminal coding65.2%65.8%61.5%Opus 5.5
Frontier coding46.2%47.3%43.8%Opus 5.5
SWE-sweep39.0%10.1%8.3%Neon
Automation56.3%42.5%41.3%Neon
Professional agents75.8%71.6%73.0%Neon
Science70.5%63.3%68.0%Neon
Hard reasoning71.3%67.9%64.0%Neon
Long context95.2%66.8%80.1%Neon
Chart reasoning73.5%66.3%75.1%GPT-6.2 Sol
Video understanding89.0%83.7%87.6%Neon
Cyber patching70.7%67.1%66.0%Neon
API input price$3$4$2GPT-6.2 Sol
API output price$15$20$10GPT-6.2 Sol