Gemini 3.8 FlashvsGemini 3.7 Flash
在 11 个共同 benchmark 中，Gemini 3.8 Flash 整体领先：Gemini 3.8 Flash 领先 10 项，Gemini 3.7 Flash 领先 1 项，持平 0 项，平均分差 +5.83。
Gemini 3.8 Flash
Google Deep Mind · 2026-09-02 · 多模态大模型
Gemini 3.7 Flash
Google Deep Mind · 2026-08-13 · 多模态大模型
Gemini 3.8 Flash10 项(91%)(9%)1 项Gemini 3.7 Flash
评测分数
按能力类目分组，每组内按分差大小排列；共 11 项。
AI Agent - Tool Usage
Gemini 3.8 Flash 领先 5/5| 评测项 | Gemini 3.8 Flash | Gemini 3.7 Flash | 分差 |
|---|---|---|---|
| BioMysteryBench (Human-difficult) | 56.501 / 2Thinking Medium (With Tools + Internet) | 43.502 / 2Thinking Medium (With Tools + Internet) | +13 |
| OSWorld 2.0 | 594 / 7Thinking (With Tools) | 47.907 / 7Thinking Medium (With Tools) | +11.10 |
| LABBench2 | 86.201 / 2Thinking Medium (With Tools + Internet) | 82.102 / 2Thinking Medium (With Tools + Internet) | +4.10 |
| Terminal-Bench 2.1 | 89.401 / 48Thinking (With Tools) | 85.8010 / 48Thinking (With Tools) | +3.60 |
| BioMysteryBench (Human-solvable) | 88.801 / 2Thinking Medium (With Tools + Internet) | 87.102 / 2Thinking Medium (With Tools + Internet) | +1.70 |
Multimodal Understanding
Gemini 3.8 Flash 领先 2/3| 评测项 | Gemini 3.8 Flash | Gemini 3.7 Flash | 分差 |
|---|---|---|---|
| CharXiv RQ | 86.208 / 19Thinking Medium (No Tools) | 88.706 / 19Thinking Medium (With Tools) | -2.50 |
| LVBench | 87.801 / 4Thinking Medium (With Tools) | 85.403 / 4Thinking Medium (No Tools) | +2.40 |
| GDP.pdf | 351 / 2Thinking Medium (No Tools) | 342 / 2Thinking Medium (No Tools) | +1 |
Coding and Software Engineer
Gemini 3.8 Flash 领先 1/1| 评测项 | Gemini 3.8 Flash | Gemini 3.7 Flash | 分差 |
|---|---|---|---|
| DeepSWE | 73.701 / 33Thinking (With Tools) | 65.3012 / 33Thinking High (With Tools) | +8.40 |
General Knowledge
Gemini 3.8 Flash 领先 1/1| 评测项 | Gemini 3.8 Flash | Gemini 3.7 Flash | 分差 |
|---|---|---|---|
| HLE-Verified | 54.901 / 2Thinking Medium (No Tools) | 53.602 / 2Thinking Medium (No Tools) | +1.30 |
Productivity Knowledge
Gemini 3.8 Flash 领先 1/1| 评测项 | Gemini 3.8 Flash | Gemini 3.7 Flash | 分差 |
|---|---|---|---|
| GDPval-AA v2 | 1,54513 / 24Thinking (No Tools) | 1,52515 / 24Thinking (No Tools) | +20 |
规格对比
| 字段 | Gemini 3.8 Flash | Gemini 3.7 Flash |
|---|---|---|
| 发布机构 | Google Deep Mind | Google Deep Mind |
| 发布时间 | 2026-09-02 | 2026-08-13 |
| 模型类型 | 多模态大模型 | 多模态大模型 |
| 架构 | 稠密模型 | 稠密模型 |
| 参数规模 | 暂无数据 | 暂无数据 |
| 上下文长度 | 1M | 1M |
| 最大输出 | 64K | 64K |
API 调用价格
价格优先使用 DataLearner 配置的 API 记录；缺失项不做推测。
| 价格项 | Gemini 3.8 Flash | Gemini 3.7 Flash |
|---|---|---|
| 文本输入 | $0.75 / 1M tokens | $0.75 / 1M tokens |
| 文本输出 | $3.75 / 1M tokens | $3.75 / 1M tokens |
| 缓存读取 | $0.075 / 1M tokens | $0.075 / 1M tokens |
小结
- Gemini 3.8 Flash在以下类目领先:AI Agent - Tool Usage (5/5)、Multimodal Understanding (2/3)、Coding and Software Engineer (1/1)、General Knowledge (1/1)、Productivity Knowledge (1/1)
11 个共同 benchmark 上，Gemini 3.8 Flash 平均高出 5.83 分。
单项差距最大的 benchmark：GDPval-AA v2 — Gemini 3.8 Flash 1,545，Gemini 3.7 Flash 1,525（分差 +20）。
本页正文由结构化模型、价格与 benchmark 数据生成，不使用实时 LLM 撰写。