Gemini 3.8 Flash vs GPT-6 Sol: Benchmark Comparison
Gemini 3.8 Flash has the higher score on 9 of 21 shared benchmarks; GPT-6 Sol leads on 12.
The largest observed score gap is 35.00 pts on ProofBench v1.1 , where GPT-6 Sol leads.
Reported ±1 standard-error ranges overlap on 5 of 20 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 | Gemini 3.8 Flash | GPT-6 Sol | Gap | Reported uncertainty |
|---|---|---|---|---|
| Vals Index | 54.83% ±1.03 | 57.54% ±1.01 | 2.71 pts | Reported ±1 SE ranges do not overlap |
| Vals RSI Index | 19.93% | 28.11% | 8.18 pts | Uncertainty comparison unavailable |
| Harvey's Legal Agent Benchmark | 10.00% ±2.42 | 1.67% ±0.82 | 8.33 pts | Reported ±1 SE ranges do not overlap |
| Legal Research Bench | 38.94% ±3.39 | 28.85% ±3.15 | 10.10 pts | Reported ±1 SE ranges do not overlap |
| EMB | 72.20% ±2.42 | 71.53% ±2.35 | 0.67 pts | Reported ±1 SE ranges overlap |
| Finance Agent (v2) | 61.44% ±0.13 | 49.05% ±0.58 | 12.39 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 66.77% ±3.15 | 53.05% ±3.29 | 13.73 pts | Reported ±1 SE ranges do not overlap |
| MedCode | 48.13% ±2.18 | 47.07% ±2.12 | 1.06 pts | Reported ±1 SE ranges overlap |
| MedScribe | 84.50% ±1.94 | 82.03% ±1.94 | 2.46 pts | Reported ±1 SE ranges overlap |
| ProofBench v1.1 | 48.00% ±5.02 | 83.00% ±3.77 | 35.00 pts | Reported ±1 SE ranges do not overlap |
| BioMysteryBench | 62.22% ±1.70 | 74.81% ±2.43 | 12.59 pts | Reported ±1 SE ranges do not overlap |
| MysteryMechanism | 36.49% ±3.24 | 30.18% ±3.09 | 6.31 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench Science | 8.57% ±3.37 | 30.00% ±5.52 | 21.43 pts | Reported ±1 SE ranges do not overlap |
| SAGE | 35.06% ±3.36 | 44.79% ±3.36 | 9.73 pts | Reported ±1 SE ranges do not overlap |
| Code Migration | 36.55% ±4.18 | 57.20% ±4.21 | 20.65 pts | Reported ±1 SE ranges do not overlap |
| IOI | 56.94% ±2.26 | 82.61% ±9.24 | 25.67 pts | Reported ±1 SE ranges do not overlap |
| ProgramBench | 1.00% ±0.70 | 2.00% ±0.99 | 1.00 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench 4.0 | 19.19% ±2.52 | 44.44% ±3.54 | 25.25 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 78.65% ±3.88 | 87.82% ±2.53 | 9.17 pts | Reported ±1 SE ranges do not overlap |
| CyberBench v1.1 | 43.75% ±2.21 | 77.98% ±5.11 | 34.23 pts | Reported ±1 SE ranges do not overlap |
| Public Benefits Bench v1.1 | 65.29% ±1.24 | 56.63% ±1.29 | 8.66 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | Gemini 3.8 Flash average | GPT-6 Sol average |
|---|---|---|
| Legal | 24.47% | 15.26% |
| Finance | 66.80% | 57.88% |
| Healthcare | 66.32% | 64.55% |
| Math | 48.00% | 83.00% |
| Science | 35.76% | 45.00% |
| Education | 35.06% | 44.79% |
| Coding | 38.47% | 54.81% |
| Cyber | 43.75% | 77.98% |
| Social Mobility | 65.29% | 56.63% |
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 | Gemini 3.8 Flash cost | GPT-6 Sol cost | Gemini 3.8 Flash latency | GPT-6 Sol latency |
|---|---|---|---|---|
| Vals Index | $5.73 | $7.58 | 51m57s | 29m11s |
| Vals RSI Index | $332.04 | $665.11 | 90h00m | 90h00m |
| Harvey's Legal Agent Benchmark | $3.66 | $3.36 | 29m21s | 13m40s |
| Legal Research Bench | $1.63 | $4.90 | 5m24s | 24m09s |
| EMB | $8.24 | $1.28 | 12m49s | 10m22s |
| Finance Agent (v2) | $2.00 | $2.12 | 3m21s | 9m00s |
| Tax Agent Bench | $0.86 | $2.28 | 2m58s | 17m32s |
| MedCode | N/A | N/A | 42.89s | 62.71s |
| MedScribe | N/A | N/A | 21.14s | 81.14s |
| ProofBench v1.1 | $0.60 | $0.48 | 6m09s | 5m05s |
| BioMysteryBench | $1.55 | $0.66 | 5m27s | 5m24s |
| MysteryMechanism | $1.73 | $0.46 | 5m22s | 6m13s |
| Terminal-Bench Science | $5.64 | $5.82 | 55m11s | 1h22m |
| SAGE | N/A | N/A | 28.64s | 31.42s |
| Code Migration | $18.49 | $15.68 | 2h37m | 1h26m |
| IOI | $3.98 | $2.70 | 14m20s | 30m38s |
| ProgramBench | $10.93 | $11.67 | 46m39s | 46m11s |
| Terminal-Bench 4.0 | $8.77 | $5.79 | 1h48m | 34m56s |
| Vibe Code Bench v1.1 | $6.87 | $26.36 | 8m39s | 38m56s |
| CyberBench v1.1 | $1.31 | $1.34 | 9m49s | 12m14s |
| Public Benefits Bench v1.1 | $1.01 | $2.49 | 3m21s | 31m02s |
Results available only for Gemini 3.8 Flash
- LegalBench
- MortgageTax
- TaxEval v2
- GPQA Diamond
- MMLU Pro
- MMMU Pro
- LiveCodeBench
- SkillsBench
- SWE-bench
- Vibe Code Bench 1-100
- CUA-bench
- Time Horizon Index: KSP
Results available only for GPT-6 Sol
None.