Skip to content

Gemini 4 Argon vs Claude Opus 5.5

Google has officially announced Gemini 4 Argon. Here is how it compares with Claude Opus 5.5. Gemini 4 Argon is cheaper than Claude, but Opus 5.5 leads in many benchmarks.

Benchmarks

BenchmarkGemini 4 ArgonClaude Opus 5.5Difference
Vals Index68.9%67.0%Argon +1.9 points
AutomationBench51.3%42.5%Argon +8.8 points
Vals Finance Agent v265.4%58.6%Argon +6.8 points
Harvey Legal Agent19.6%3.8%Argon +15.8 points
DeepSWE v1.177.9%74.2%Argon +3.7 points
FrontierSWE v255.0%62.3%Opus +7.3 points
Vibe Code Bench91.9%90.3%Argon +1.6 points
Terminal-Bench 4.057.4%66.4%Opus +9.0 points
PostTrainBench45.3%49.3%Opus +4.0 points
Terminal-Bench Science 0.157.6%63.3%Opus +5.7 points
LABBench 288.8%73.1%Argon +15.7 points
RiemannBench76.0%69.6%Argon +6.4 points
GraphWalks up to 128K99.7%90.6%Argon +9.1 points
GraphWalks 256K to 1M84.2%66.8%Argon +17.4 points
Agent's Last Exam39.5%38.2%Argon +1.3 points
Chartography71.6%66.3%Argon +5.3 points
LVBench91.7%83.7%Argon +8.0 points
CWE-bench v168.0%67.0%Argon +1.0 point

Artificial Analysis benchmarks

BenchmarkGemini 4 Argon (High)Claude Opus 5.5 (Max)Difference
Intelligence Index v4.3.25358Opus +5
AA-Briefcase v1.11488 Elo1805 EloOpus +317 Elo
GDPval-AA v2.11624 Elo1866 EloOpus +242 Elo
AutomationBench-AA78%70%Argon +8 points
Terminal-Bench 4.057%60%Opus +3 points
SciCode62%67%Opus +5 points
Humanity's Last Exam57%61%Opus +4 points
GDP.pdf22%26%Opus +4 points
CritPt27%32%Opus +5 points
AA-Omniscience4246Opus +4
AA-LCR v1.180%85%Opus +5 points

Cost and efficiency

MetricGemini 4 Argon (High)Claude Opus 5.5 (Max)Difference
Intelligence Index5358Opus +5
Cost per Intelligence Index task$1.99$5.98Argon about 67% cheaper
Cost to run the Intelligence Index$2,407$8,708Argon about 72% cheaper
Output tokens per task62K119KArgon uses about 48% fewer
Reasoning tokens per task36K84KArgon uses about 57% fewer
Output tokens across the Index113M260MArgon uses about 57% fewer
Input, per 1M tokens$2$4Argon 50% cheaper
Output, per 1M tokens$10$20Argon 50% cheaper
Cached input, per 1M tokens$0.10$0.20Argon 50% cheaper

Pricing and specs

SpecificationGemini 4 ArgonClaude Opus 5.5Difference
Input now, per 1M tokens$2Introductory$4Argon 50% cheaper
Output now, per 1M tokens$10Introductory$20Argon 50% cheaper
Cached input now, per 1M tokens$0.10$0.20Argon 50% cheaper
Input later, per 1M tokens$4$4Same
Output later, per 1M tokens$20$20Same
Context window1M tokens1M tokensSame
Max output1M tokens128K tokensArgon about 7.8x larger
ReasoningYesAdaptive, always onBoth
Default effortMedium, tested at HighMediumArgon tested at High
Text inputYesYesSame
Image inputYesYesSame
AvailabilityAvailable, paid API tierAvailableBoth available