Claude Opus 4.8 vs Gemini 3.8 Flash: Benchmark Comparison
Claude Opus 4.8 has the higher score on 14 of 26 shared benchmarks; Gemini 3.8 Flash leads on 11.
The largest observed score gap is 19.72 pts on SAGE , where Claude Opus 4.8 leads.
Reported ±1 standard-error ranges overlap on 13 of 25 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 Opus 4.8 | Gemini 3.8 Flash | Gap | Reported uncertainty |
|---|---|---|---|---|
| Vals Index | 55.10% ±1.01 | 54.83% ±1.03 | 0.27 pts | Reported ±1 SE ranges overlap |
| Vals RSI Index | 23.98% | 19.93% | 4.05 pts | Uncertainty comparison unavailable |
| Harvey's Legal Agent Benchmark | 9.58% ±2.15 | 10.00% ±2.42 | 0.42 pts | Reported ±1 SE ranges overlap |
| Legal Research Bench | 43.75% ±3.45 | 38.94% ±3.39 | 4.81 pts | Reported ±1 SE ranges overlap |
| LegalBench | 83.57% ±0.47 | 86.99% ±0.43 | 3.42 pts | Reported ±1 SE ranges do not overlap |
| EMB | 69.37% ±2.61 | 72.20% ±2.42 | 2.82 pts | Reported ±1 SE ranges overlap |
| Finance Agent (v2) | 53.92% ±0.16 | 61.44% ±0.13 | 7.52 pts | Reported ±1 SE ranges do not overlap |
| MortgageTax | 69.91% ±0.89 | 65.34% ±0.94 | 4.57 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 64.91% ±3.17 | 66.77% ±3.15 | 1.86 pts | Reported ±1 SE ranges overlap |
| TaxEval v2 | 75.63% ±0.84 | 74.45% ±0.85 | 1.19 pts | Reported ±1 SE ranges overlap |
| MedCode | 53.22% ±2.17 | 48.13% ±2.18 | 5.08 pts | Reported ±1 SE ranges do not overlap |
| MedScribe | 85.75% ±1.93 | 84.50% ±1.94 | 1.26 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench Science | 4.29% ±2.44 | 8.57% ±3.37 | 4.29 pts | Reported ±1 SE ranges overlap |
| GPQA Diamond | 92.42% ±1.98 | 94.44% ±1.48 | 2.02 pts | Reported ±1 SE ranges overlap |
| MMLU Pro | 89.58% ±0.30 | 90.22% ±0.29 | 0.64 pts | Reported ±1 SE ranges do not overlap |
| MMMU Pro | 86.59% ±0.82 | 89.08% ±0.75 | 2.48 pts | Reported ±1 SE ranges do not overlap |
| SAGE | 54.79% ±3.35 | 35.06% ±3.36 | 19.72 pts | Reported ±1 SE ranges do not overlap |
| Code Migration | 47.25% ±4.18 | 36.55% ±4.18 | 10.70 pts | Reported ±1 SE ranges do not overlap |
| LiveCodeBench | 87.82% ±0.95 | 89.48% ±0.90 | 1.66 pts | Reported ±1 SE ranges overlap |
| ProgramBench | 1.00% ±0.70 | 1.00% ±0.70 | 0.00 pts | Reported ±1 SE ranges overlap |
| SkillsBench | 59.23% ±4.50 | 57.98% ±4.42 | 1.24 pts | Reported ±1 SE ranges overlap |
| SWE-bench | 88.60% ±1.42 | 80.00% ±1.79 | 8.60 pts | Reported ±1 SE ranges do not overlap |
| Terminal-Bench 4.0 | 23.23% ±1.34 | 19.19% ±2.52 | 4.04 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 82.72% ±3.08 | 78.65% ±3.88 | 4.07 pts | Reported ±1 SE ranges overlap |
| Public Benefits Bench v1.1 | 68.13% ±1.21 | 65.29% ±1.24 | 2.84 pts | Reported ±1 SE ranges do not overlap |
| Time Horizon Index: KSP | 11.83% ±0.00 | 18.83% ±0.00 | 7.00 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | Claude Opus 4.8 average | Gemini 3.8 Flash average |
|---|---|---|
| Legal | 45.63% | 45.31% |
| Finance | 66.75% | 68.04% |
| Healthcare | 69.49% | 66.32% |
| Science | 4.29% | 8.57% |
| Academic | 89.53% | 91.25% |
| Education | 54.79% | 35.06% |
| Coding | 55.69% | 51.84% |
| Social Mobility | 68.13% | 65.29% |
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 Opus 4.8 cost | Gemini 3.8 Flash cost | Claude Opus 4.8 latency | Gemini 3.8 Flash latency |
|---|---|---|---|---|
| Vals Index | $13.14 | $5.73 | 39m48s | 51m57s |
| Vals RSI Index | $465.25 | $332.04 | 90h00m | 90h00m |
| Harvey's Legal Agent Benchmark | $10.22 | $3.66 | 24m08s | 29m21s |
| Legal Research Bench | $2.82 | $1.63 | 11m54s | 5m24s |
| LegalBench | N/A | N/A | 3.11s | 3.32s |
| EMB | $12.06 | $8.24 | 43m52s | 12m49s |
| Finance Agent (v2) | $4.22 | $2.00 | 9m07s | 3m21s |
| MortgageTax | N/A | N/A | 20.81s | 14.79s |
| Tax Agent Bench | $1.69 | $0.86 | 7m59s | 2m58s |
| TaxEval v2 | N/A | N/A | 70.33s | 8.76s |
| MedCode | N/A | N/A | 105.92s | 42.89s |
| MedScribe | N/A | N/A | 82.36s | 21.14s |
| Terminal-Bench Science | $23.14 | $5.64 | 1h55m | 55m11s |
| GPQA Diamond | N/A | N/A | 2m14s | 20.18s |
| MMLU Pro | N/A | N/A | 23.44s | 7.82s |
| MMMU Pro | N/A | N/A | 49.04s | 12.03s |
| SAGE | N/A | N/A | 3m06s | 28.64s |
| Code Migration | $30.51 | $18.49 | 1h15m | 2h37m |
| LiveCodeBench | N/A | N/A | 2m40s | 24.07s |
| ProgramBench | $31.27 | $10.93 | 1h32m | 46m39s |
| SkillsBench | $4.77 | $2.34 | 15m30s | 3m55s |
| SWE-bench | $1.92 | $2.19 | 9m27s | 11m23s |
| Terminal-Bench 4.0 | $17.14 | $8.77 | 1h11m | 1h48m |
| Vibe Code Bench v1.1 | $26.88 | $6.87 | 1h16m | 8m39s |
| Public Benefits Bench v1.1 | $1.89 | $1.01 | 12m48s | 3m21s |
| Time Horizon Index: KSP | $1499.32 | $543.14 | N/A | N/A |
Results available only for Claude Opus 4.8
None.
Results available only for Gemini 3.8 Flash
- ProofBench v1.1
- BioMysteryBench
- MysteryMechanism
- IOI
- Vibe Code Bench 1-100
- CyberBench v1.1
- CUA-bench