Google: gemini-4-argon
25 benchmarks
Leaderboard position
#9
Overall rank (of 1121 models)
3129
Elo rating
25
Benchmarks
Top 0.8%
Percentile
In Google (of 72 models)
Above: Moonshot AI: kimi-k3.1-pro #8
Below: OpenAI: gpt-6-astra #10
Benchmark highlights
Best benchmark
VibeCodeBench
91.9
Worst benchmark
BioReason Coding
6.5
Nearest competitor from another company
The model from another company with the most similar benchmark scores.
appeared as nearest in 10 of 25 benchmarks
10/25
40.0%
| Model | Votes | % of benchmarks |
|---|---|---|
| Anthropic: claude-opus-5.5 | 10 | 40.0% |
| OpenAI: gpt-6.1-sol | 3 | 12.0% |
| OpenAI: gpt-6-astra | 3 | 12.0% |
| Anthropic: claude-fable-5.2 | 2 | 8.0% |
| Anthropic: claude-fable-5.1 | 2 | 8.0% |
| huggingfacebio: carbon-x | 2 | 8.0% |
| DeepSeek: deepseek-v4.1-flash-1017 | 1 | 4.0% |
| Qwen: qwen3.8-max | 1 | 4.0% |
| Meta: muse-wonder-full | 1 | 4.0% |
Benchmarks
VibeCodeBench
91.9
LVBench
91.7
BioReason Non-SNV
90.3
LabBench2
88.8
DeepSWE 1.1
77.9
RiemannBench
76
Chartography
71.6
Vals Index
68.9
CWE-Bench v1
68
Vals Finance Agent 2
65.4
GenomeQA
63.1
frontiervep
58
Terminal-Bench-Science 0.1
57.6
Terminal-Bench v4.0
57.4
FrontierSWE 2
55
AA-Robotics v0.1
55
Artificial Analysis Intelligence Index v4.3.2
53
AutomationBench
51.3
Artifical Analysis Intelligence Index v5.0
50
PostTrainBench
45.3
ExploitGym
42
Agents' Last Exam (Pass Rate) (Official Agent)
39.5
NextBase
31.2
Harvey's Legal Agent Benchmark
19.6
BioReason Coding
6.5
Predicted scores
Estimated from 25 similar models and the global Elo rating. These are projections, not measured results.