Model dossier模型档案
Mistral Medium 3.5
Mistral Medium 3.5 is a mid-tier model from Mistral AI (Intelligence Score 64/100). Strongest in agent (95.1) and math (86.3); best fit: tool-use and multi-step agents. It trails in preference (32). Premium-priced: $1.50 in / $7.50 out per 1M tokens. Open weights — self-hostable.
Mistral Medium 3.5 是 Mistral AI 的中端梯队模型(智能评分 64/100)。 最强项是智能体(95.1分),其次是数学(86.3分);适合工具调用与多步 Agent。 短板是偏好(32分)。 定价属高端定价:每百万 token 输入 $1.50 / 输出 $7.50。 开放权重,可自托管。
open weights开放权重
64
Intelligence Score · benchmark coverage 智能评分 · benchmark 覆盖 88%
◈
Benchmark scoresBenchmark 成绩
9 entries 条| Benchmark基准 | Score成绩 | Index指数 | Dated日期 | Measured by测评方 | |
|---|---|---|---|---|---|
| Humanity's Last ExamReasoning | 4.3 | 7 | — | official官方榜 | source ↗×5⚠±8.5 |
| GPQA DiamondReasoning | 74.8 | 78 | — | 3rd-party第三方 | source ↗×4 |
| MMLU-ProKnowledge | 81.3 | 86 | — | 3rd-party第三方 | source ↗×2⚠±5.3 |
| SWE-bench VerifiedCoding | 77.6 | 81 | 2026-08 | 3rd-party第三方 | source ↗×6 |
| AIME 2025Math | 86.3 | 86 | — | 3rd-party第三方 | source ↗×2⚠±46.3 |
| LiveCodeBenchCoding | 44.84 | 48 | — | 3rd-party第三方 | source ↗×2⚠±29 |
| Terminal-BenchCoding | 66 | 72 | — | 3rd-party第三方 | source ↗ |
| τ²-benchAgent | 94.2 | 95 | — | 3rd-party第三方 | source ↗×3 |
| LMArena (Chatbot Arena)Preference | 1427 | 32 | 2026-08-12 | official官方榜 | source ↗×2 |
≡
Specs & pricing规格与价格
| Provider厂商 | Mistral AI |
|---|---|
| Released发布 | 2026-04-30 |
| Context window上下文 | 262K tokens |
| Max output最大输出 | 210K tokens |
| Modality模态 | text+image+file->text |
| Reasoning tiers推理档位 | not documented未见公开文档 |
| API input price输入价格 | $1.50 / 1M tokens |
| API output price输出价格 | $7.50 / 1M tokens |
⛓
Tools using this model使用它的工具
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。