MiniMax 发布 M3.1 Flash,推理提速近三倍、用量成本降低 70%

来源:MiniMax:Blog(网页) 2026年10月18日 00:00 AIHOT 评分:70 精选
摘要:MiniMax 发布 MiniMax M3.1 Flash,匿名模型 Space Bunny 是其早期版本,曾处理超过 200 万亿 token。
推荐理由:官方给出具体基准分数、推理速度和每百万 token 定价,读者可直接对比智能体任务的性价比变化。
# MiniMax 发布 M3.1 Flash,推理提速近三倍、用量成本降低 70% > 原标题:MiniMax M3.1 Flash: Redefining Agent Performance and Cost - 来源:MiniMax:Blog - 发布时间:2026-10-18T16:00:00.000Z - AIHOT:https://aihot.news/items/wrnv8dd428iozlj5xax99zbmj - 原文:https://www.minimax.io/blog/minimax-m31f ## 摘要 MiniMax 发布 MiniMax M3.1 Flash,匿名模型 Space Bunny 是其早期版本,曾处理超过 200 万亿 token。 ## 推荐理由 官方给出具体基准分数、推理速度和每百万 token 定价,读者可直接对比智能体任务的性价比变化。 ## 正文 · 中文译文 今天,我们正式推出 **MiniMax M3.1 Flash**。 Space Bunny 🐰 在匿名运行期间处理了超过 200 万亿个 token,它是 MiniMax M3.1 Flash 的早期版本。 **作为新一代 Flash 模型,M3.1 Flash 在智能体任务、编程、Computer Use 和网络安全方面实现了代际跃升,同时将在线推理速度提升近三倍,并将使用成本降低 70%。** ### 全面增强的能力 M3.1 Flash 在广泛的基准测试中均有提升,在复杂智能体任务(cowork)、计算机使用、软件开发(编程)和网络安全方面达到领先模型水平。 ![](https://filecdn.minimax.chat/public/m31-flash-en-2-1792403273121.png) **复杂任务与自动化:** - M3.1 Flash 在 AutomationBench-AA 上位列顶尖表现者之列。它能够跨 SaaS 系统完成真实企业工作流,端到端解决销售、营销、运营、客户服务、财务和人力资源中的实际问题。 - 在 BankerToolBench 上得分 76.2,擅长应对多步规划、跨文档和文件保持一致、财务合规以及链式工具调用等复杂的现实挑战。 **编程与软件工程:** - 它在 DeepSWE v1.1 上达到 73.7%,跻身顶尖模型之列。DeepSWE 聚焦复杂软件任务,要求模型独立理解代码、开发功能、修复问题并通过测试验证工作。M3.1 Flash 在长序列步骤中展现出强大的执行能力,以更低成本支持复杂软件开发工作流。 **Computer Use:** - 它在衡量 AI 计算机操作与跨应用任务执行的 OSWorld V2 上得分 61.7 分,在评估真实专业工作和复杂任务表现的 Agents’ Last Exam(ALE)上得分 29.6 分。它在这两项基准上均位列中国模型第一。 - M3.1 Flash 在与真实环境交互、使用工具和执行长序列任务方面展现出强大能力——不止于理解任务,更能采取行动并交付结果。 **网络安全:** - 它在 CyberGym 上取得 82% 的优异成绩,在漏洞发现与验证方面展现出领先能力。 - 在我们内部安全专家的深度评估中,该模型在代码审计、漏洞修补、渗透测试、恶意软件分析和逆向工程方面表现强劲,能够高效、低成本地支持构建复杂安全系统。 我们通过更激进的稀疏化进一步优化了 M3.1 Flash 的推理效率。在日常使用中,输出速度可达 200+ TPS(每秒 token 数),中位数(P50)速度稳定达到 170+ TPS。 ### 定价:最具成本效益 对于智能体任务,该模型的整体使用成本比上一代降低 70%。 **每 1M token 统一价格:输入 $0.15 · 输出 $0.60 · 缓存输入 $0.015。** | 模型 | 输入
美元 / 1M token | 输出
美元 / 1M token | 缓存输入
美元 / 1M token | | --- | --- | --- | --- | | MiniMax-
M3.1-Flash | $0.15 | $0.60 | $0.015 | ### 重新定义智能体任务的能力与成本 在三个基准测试中——用于复杂智能体任务执行的 Automation-Bench、用于真实世界编程的 DeepSWE,以及用于评估具有经济价值的真实世界任务中计算机使用能力的 Agents’ Last Exam(ALE)——M3.1 Flash 以极低的任务成本实现了与顶级模型相当的性能,为性能与成本树立了极具竞争力的新标准。 ![](https://filecdn.minimax.chat/public/m31-flash-en-5-1792403273121.png) ![](https://filecdn.minimax.chat/public/m31-flash-en-4-1792403273121.png) ![](https://filecdn.minimax.chat/public/m31-flash-en-3-1792403273121.png) **M3.1 Flash 现已通过 MiniMax API、M Plan 和 MiniMax Code 提供。模型权重将于稍后发布。** 下载桌面版 MiniMax Code:[https://agent.minimax.io/download](https://agent.minimax.io/download) 订阅 M Plan:[https://www.minimax.io/m-plan](https://www.minimax.io/m-plan) API 访问:[platform.minimax.io](https://t.co/fHRdSV73Hr) Flash 代表了我们持续探索效率与能力之间平衡的努力。旨在突破智能边界的 Pro 也将于稍后推出。 让智能惠及每一个人。 ## 正文 · 原文 Today, we officially roll out **MiniMax M3.1 Flash**. Space Bunny 🐰, which processed more than 200 trillion tokens while operating anonymously, was an early version of MiniMax M3.1 Flash. **As a next-generation Flash model, M3.1 Flash delivers a generational leap in agentic tasks, coding, Computer Use and cybersecurity, while making online inference nearly three times faster and reducing usage costs by 70%.** ### Stronger Capabilities Across the Board M3.1 Flash improves across a broad range of benchmarks, reaching the level of leading models in complex agentic tasks (cowork), computer use, software development (coding), and cybersecurity. ![](https://filecdn.minimax.chat/public/m31-flash-en-2-1792403273121.png) **Complex Tasks and Automation:** - M3.1 Flash ranks among the top performers on AutomationBench-AA. It can complete real enterprise workflows across SaaS systems, solving practical problems in sales, marketing, operations, customer service, finance, and HR from end to end. - Scoring 76.2 on BankerToolBench, it excels at complex real-world challenges such as multi-step planning, maintaining consistency across documents and files, financial compliance, and chained tool calls. **Coding and Software Engineering:** - It achieves 73.7% on DeepSWE v1.1, placing it alongside top models. DeepSWE focuses on complex software tasks that require models to independently understand code, develop features, fix issues, and validate their work through testing. M3.1 Flash demonstrates strong execution across long sequences of steps, supporting complex software development workflows at a lower cost. **Computer Use:** - It scores 61.7 points on OSWorld V2, which measures AI computer operation and cross-application task execution, and 29.6 points on Agents’ Last Exam (ALE), which evaluates performance on real professional work and complex tasks. It ranks first among Chinese models on both benchmarks. - M3.1 Flash demonstrates strong capabilities in interacting with real environments, using tools, and executing long sequences of tasks—moving beyond understanding a task to taking action and delivering results. **Cybersecurity:** - It achieves a strong score of 82% on CyberGym, demonstrating leading capabilities in vulnerability discovery and validation. - In in-depth evaluations by our internal security experts, the model performed strongly in code auditing, vulnerability patching, penetration testing, malware analysis, and reverse engineering, enabling efficient, low-cost support for building complex security systems. We have further optimized M3.1 Flash’s inference efficiency through more aggressive sparsity. In everyday use, output speeds can reach 200+ TPS (tokens per second), with median (P50) speeds consistently reaching 170+ TPS. ### Pricing: Most Cost-Effective For agentic tasks, the model’s overall usage cost is 70% lower than its predecessor. **Flat pricing per 1M tokens: $0.15 input · $0.60 output · $0.015 cached input.** | Model | Input
USD / 1M tokens | Output
USD / 1M tokens | Cached input
USD / 1M tokens | | --- | --- | --- | --- | | MiniMax-
M3.1-Flash | $0.15 | $0.60 | $0.015 | ### Redefining Capability and Cost for Agentic Tasks Across three benchmarks—Automation-Bench for complex agentic task execution, DeepSWE for real-world coding, and Agents’ Last Exam (ALE) for computer use in real-world tasks with economic value—M3.1 Flash delivers performance on par with top models at a very low task cost, setting a highly competitive new standard for performance and cost. ![](https://filecdn.minimax.chat/public/m31-flash-en-5-1792403273121.png) ![](https://filecdn.minimax.chat/public/m31-flash-en-4-1792403273121.png) ![](https://filecdn.minimax.chat/public/m31-flash-en-3-1792403273121.png) **M3.1 Flash is now available through the MiniMax API, M Plan, and MiniMax Code. Model weights will be released later.** Download MiniMax Code for desktop: [https://agent.minimax.io/download](https://agent.minimax.io/download) Subscribe to M Plan: [https://www.minimax.io/m-plan](https://www.minimax.io/m-plan) API access: [platform.minimax.io](https://t.co/fHRdSV73Hr) Flash represents our ongoing exploration of the balance between efficiency and capability. Pro, which pushes the boundaries of intelligence, will also arrive later. Intelligence with Everyone.