Claude Sonnet 5.5 vs Gemini 4 Argon: Benchmark Comparison
Claude Sonnet 5.5 has the higher score on 16 of 28 shared benchmarks; Gemini 4 Argon leads on 12.
The largest observed score gap is 18.27 pts on CyberBench v1.1 , where Gemini 4 Argon leads.
Reported ±1 standard-error ranges overlap on 15 of 28 shared benchmarks with comparable uncertainty data. This is not a pairwise statistical significance test.
Shared benchmark results
Scores and reported standard errors come from the latest visible Vals benchmark version available for both models.
| Benchmark | Claude Sonnet 5.5 | Gemini 4 Argon | Gap | Reported uncertainty |
|---|---|---|---|---|
| Vals Index | 70.85% ±0.93 | 69.39% ±1.09 | 1.47 pts | Reported ±1 SE ranges overlap |
| Vals Finance Index | 64.53% ±1.48 | 66.00% ±1.49 | 1.47 pts | Reported ±1 SE ranges overlap |
| Vals Legal Index | 68.02% ±1.81 | 76.78% ±1.84 | 8.76 pts | Reported ±1 SE ranges do not overlap |
| Vals Science Index | 59.12% ±2.31 | 55.36% ±2.29 | 3.76 pts | Reported ±1 SE ranges overlap |
| HLAB-Vals-1.1 | 10.42% ±2.57 | 27.08% ±3.74 | 16.67 pts | Reported ±1 SE ranges do not overlap |
| Immigration | 68.96% ±1.54 | 68.08% ±1.45 | 0.89 pts | Reported ±1 SE ranges overlap |
| Legal Research Bench | 48.08% ±3.47 | 54.81% ±3.46 | 6.73 pts | Reported ±1 SE ranges overlap |
| EMB v1.1 | 75.90% ±2.47 | 74.83% ±2.49 | 1.07 pts | Reported ±1 SE ranges overlap |
| Finance Agent v2.1 | 67.28% ±2.02 | 67.66% ±2.00 | 0.38 pts | Reported ±1 SE ranges overlap |
| Tax Agent Bench | 41.96% ±3.56 | 49.22% ±3.61 | 7.26 pts | Reported ±1 SE ranges do not overlap |
| MedCode | 52.92% ±2.12 | 58.80% ±2.10 | 5.88 pts | Reported ±1 SE ranges do not overlap |
| MedScribe | 91.10% ±1.96 | 87.43% ±1.96 | 3.67 pts | Reported ±1 SE ranges overlap |
| ProofBench v1.1 | 100.00% ±0.00 | 99.00% ±1.00 | 1.00 pts | Reported ±1 SE ranges overlap |
| BioMysteryBench | 81.11% ±0.64 | 76.30% ±0.37 | 4.81 pts | Reported ±1 SE ranges do not overlap |
| MysteryMechanism | 49.10% ±3.36 | 45.49% ±3.35 | 3.60 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench-Science | 47.14% ±6.01 | 44.29% ±5.98 | 2.86 pts | Reported ±1 SE ranges overlap |
| SAGE | 88.03% ±1.62 | 89.46% ±1.64 | 1.43 pts | Reported ±1 SE ranges overlap |
| Code Migration v1.1 | 86.88% ±1.52 | 70.80% ±3.72 | 16.08 pts | Reported ±1 SE ranges do not overlap |
| IOI | 83.06% ±3.74 | 100.00% ±0.00 | 16.94 pts | Reported ±1 SE ranges do not overlap |
| ProgramBench | 6.50% ±1.75 | 2.50% ±1.11 | 4.00 pts | Reported ±1 SE ranges do not overlap |
| SWE-bench Multimodal | 58.33% ±2.25 | 56.04% ±2.27 | 2.29 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench 4.0 | 64.14% ±1.01 | 57.58% ±2.31 | 6.57 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 92.39% ±1.26 | 91.91% ±1.90 | 0.49 pts | Reported ±1 SE ranges overlap |
| CyberBench v1.1 | 59.58% ±5.21 | 77.86% ±5.30 | 18.27 pts | Reported ±1 SE ranges do not overlap |
| SRE Bench | 30.15% ±2.84 | 44.27% ±3.08 | 14.12 pts | Reported ±1 SE ranges do not overlap |
| Access to Justice Benchmark | 74.94% ±0.70 | 68.48% ±0.75 | 6.46 pts | Reported ±1 SE ranges do not overlap |
| Financial Coaching Bench | 74.94% ±1.55 | 73.13% ±1.47 | 1.81 pts | Reported ±1 SE ranges overlap |
| Public Benefits Bench v1.1 | 67.19% ±1.22 | 69.76% ±0.00 | 2.57 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | Claude Sonnet 5.5 average | Gemini 4 Argon average |
|---|---|---|
| Legal | 42.49% | 49.99% |
| Finance | 61.71% | 63.90% |
| Healthcare | 72.01% | 73.11% |
| Math | 100.00% | 99.00% |
| Science | 59.12% | 55.36% |
| Education | 88.03% | 89.46% |
| Coding | 65.22% | 63.14% |
| Cyber | 44.87% | 61.07% |
| Social Mobility | 72.36% | 70.46% |
Cost and latency
Cost per test appears only where the benchmark reports it for both models. Latency is the measured completion time for that benchmark.
| Benchmark | Claude Sonnet 5.5 cost | Gemini 4 Argon cost | Claude Sonnet 5.5 latency | Gemini 4 Argon latency |
|---|---|---|---|---|
| Vals Index | $16.18 | $13.37 | 1h26m | 44m58s |
| Vals Finance Index | $7.40 | $6.53 | 43m31s | 22m41s |
| Vals Legal Index | $13.37 | $6.82 | 1h15m | 25m41s |
| Vals Science Index | $8.50 | $18.40 | 1h09m | 57m00s |
| HLAB-Vals-1.1 | $17.71 | $10.55 | 1h13m | 29m15s |
| Immigration | $1.16 | $0.51 | 11m03s | 5m09s |
| Legal Research Bench | $13.66 | $6.45 | 1h33m | 31m21s |
| EMB v1.1 | $6.53 | $11.80 | 36m45s | 38m26s |
| Finance Agent v2.1 | $6.95 | $4.32 | 34m26s | 13m13s |
| Tax Agent Bench | $8.73 | $3.47 | 59m21s | 16m26s |
| MedCode | N/A | N/A | 3m59s | 79.53s |
| MedScribe | N/A | N/A | 5m07s | 87.44s |
| ProofBench v1.1 | $0.52 | $0.78 | 5m48s | 6m20s |
| BioMysteryBench | $2.95 | $5.71 | 18m06s | 39m57s |
| MysteryMechanism | $2.98 | $4.59 | 23m30s | 24m55s |
| Terminal-Bench-Science | $19.57 | $44.90 | 2h46m | 1h46m |
| SAGE | $0.16 | $0.07 | 81.12s | 29.17s |
| Code Migration v1.1 | $43.88 | $44.79 | 4h07m | 2h22m |
| IOI | $5.81 | $5.31 | 45m51s | 24m19s |
| ProgramBench | $22.62 | $17.66 | 2h07m | 1h11m |
| SWE-bench Multimodal | $3.26 | $11.40 | 24m13s | 33m38s |
| Terminal-Bench 4.0 | $10.75 | $17.64 | 1h22m | 53m35s |
| Vibe Code Bench v1.1 | $19.73 | $8.11 | 1h01m | 37m28s |
| CyberBench v1.1 | $2.91 | $5.11 | 33m53s | 19m10s |
| SRE Bench | $17.27 | $30.45 | 1h55m | 1h35m |
| Access to Justice Benchmark | $2.37 | $0.83 | 17m34s | 6m05s |
| Financial Coaching Bench | $0.80 | $0.16 | 8m45s | 4m48s |
| Public Benefits Bench v1.1 | $3.76 | $2.79 | 1h15m | 26m07s |
Results available only for Claude Sonnet 5.5
- ClaimBench
Results available only for Gemini 4 Argon
- Vals RSI Index
- LegalBench
- Autoform-Hard
- Robo-bench
- CUA-bench
- Time Horizon Index: KSP