Model dossier模型档案
Gemini 3.1 Pro Preview
Gemini 3.1 Pro Preview is a high-tier model from Google (Intelligence Score 82.4/100). Strongest in knowledge (94.5) and reasoning (90.5); best fit: knowledge QA and long-form writing. It trails in preference (46.8). Premium-priced: $2.00 in / $12 out per 1M tokens. Context window 1M tokens — long-document friendly.
Gemini 3.1 Pro Preview 是 Google 的高水平梯队模型(智能评分 82.4/100)。 最强项是知识(94.5分),其次是推理(90.5分);适合知识问答与长文写作。 短板是偏好(46.8分)。 定价属高端定价:每百万 token 输入 $2.00 / 输出 $12。 上下文 1M token,长文档友好。
82.4
Intelligence Score · benchmark coverage 智能评分 · benchmark 覆盖 100%
◈
Benchmark scoresBenchmark 成绩
11 entries 条| Benchmark基准 | Score成绩 | Index指数 | Dated日期 | Measured by测评方 | |
|---|---|---|---|---|---|
| Humanity's Last ExamReasoning | 46.44⚡high | 81 | — | official官方榜 | source ↗×9⚠±5.94 |
| GPQA DiamondReasoning | 95.5 | 100 | — | 3rd-party第三方 | source ↗×9 |
| MMLU-ProKnowledge | 89.8 | 95 | 2026-08-14 | 3rd-party第三方 | source ↗×5 |
| SWE-bench VerifiedCoding | 80.6 | 84 | 2026-08 | 3rd-party第三方 | source ↗×9⚠±3.3 |
| AIME 2025Math | 93.3 | 93 | — | 3rd-party第三方 | source ↗×5⚠±8.8 |
| LiveCodeBenchCoding | 91.7 | 98 | 2026-03 | 3rd-party第三方 | source ↗×6⚠±30.7 |
| FrontierMathMath | 16.7 | 19 | — | 3rd-party第三方 | source ↗×2⚠±20.2 |
| Terminal-BenchCoding | 68.5 | 75 | — | official官方榜 | source ↗×6⚠±41.4 |
| τ²-benchAgent | 95.6 | 96 | — | 3rd-party第三方 | source ↗×5⚠±3.7 |
| LMArena (Chatbot Arena)Preference | 1486 | 47 | 2026-08-12 | official官方榜 | source ↗×9 |
| WebArenaAgent | 69 | 75 | — | 3rd-party第三方 | source ↗ |
≡
Specs & pricing规格与价格
| Provider厂商 | |
|---|---|
| Released发布 | 2026-02-19 |
| Context window上下文 | 1M tokens |
| Max output最大输出 | 66K tokens |
| Modality模态 | text+image+file+audio+video->text |
| Reasoning tiers推理档位 | ⚡low ⚡medium ⚡high default默认 high · minimal / low / medium / high |
| API input price输入价格 | $2.00 / 1M tokens |
| API output price输出价格 | $12 / 1M tokens |
⛓
Tools using this model使用它的工具
⚔
Head-to-head正面对比
How we score →评分方法 → · Catalog & pricing via the official OpenRouter API.模型目录与价格来自 OpenRouter 官方 API。