Claude Opus 4.8 vs Gemini 3.5 Flash
Claude Opus 4.8 
Gemini 3.5 Flash Verdict
- Claude Opus 4.8 and Gemini 3.5 Flash (medium) are level on quality (64.7 vs 63.9).
- Claude Opus 4.8 is stronger in coding, composite, long context.
- Gemini 3.5 Flash (medium) is stronger in agents & tools, human preference, instruction following, knowledge, multimodal, reasoning.
- Gemini 3.5 Flash (medium) is 3.0× cheaper ($3.38 vs $10.00 per 1M blended).
- Gemini 3.5 Flash (medium) streams 3.6× faster (209 vs 58 tokens per second).
| Metric | Claude Opus 4.8 | Gemini 3.5 Flash (medium) |
|---|---|---|
| BenchLeader Index | 64.7 | 63.9 |
| Agents & tools score | 65.0 | 67.8 |
| Coding score | 65.7 | 58.9 |
| Composite score | 81.9 | 71.4 |
| Human preference score | 65.6 | 67.4 |
| Instruction following score | 61.6 | 72.2 |
| Knowledge score | 66.3 | 73.9 |
| Long context score | 65.3 | 63.6 |
| Multimodal score | 64.4 | 67.6 |
| Reasoning score | 64.2 | 66.8 |
| Blended price $/M | $10.00 | $3.38 |
| Output speed | 58 tok/s | 209 tok/s |
| Time to first answer | 29.5 s | 17.0 s |
| Context window | 1M | 1M |
| SimpleBench | 64.8% | – |
| Remote Labor Index | 8.3% | – |
| FrontierCode | 46.5% | – |
| GSO-Bench | 47.1% | – |
| Epoch Capabilities Index | 158.3 | – |
| LMArena Text | 1473 | 1476 |
| LMArena Hard Prompts | 1503 | 1493 |
| LMArena Coding | 1527 | 1503 |
| LMArena WebDev | 1540 | 1491 |
| LMArena Vision | 1289 | 1306 |
| AA Intelligence Index | 42.0 | 33.6 |
| IFBench | 62.2% | 74.6% |
| AA-LCR | 77.7% | 74.3% |
| MMMU-Pro | – | 83.9% |
| AA-Omniscience | 28.8 | 20.8 |
| Terminal-Bench Hard | 58.3% | 39.4% |
| GPQA Diamond (AA) | 92.0% | 92.1% |
| Humanity's Last Exam (AA) | 48.7% | 41.3% |
| SciCode (AA) | 54.4% | – |
| τ²-Bench Telecom (AA) | 94.4% | 95.6% |
| LiveCodeBench | 87.8% | – |
| MMLU-Pro | 89.6% | – |
| LegalBench | 83.6% | – |
| CorpFin | 66.7% | – |
| TaxEval | 75.6% | – |
| Terminal-Bench 2.1 (Vals) | 71.9% | – |
| SWE-bench (Vals) | 88.6% | – |
| GPQA Diamond (Vals) | 92.4% | – |
| Vals Index | 60.9 | – |
| HiL-Bench | 35.3% | – |
| EQ-Bench 4 | 1281 | – |
Data as of 2026-09-09. Best configuration of each model; every score links to its source on the model pages.