模型发布时间轴 - XSCT Bench

2026

2 月

Google: Gemini 3.1 Pro Preview 待评测

2026-02-19 Google

多模态代码推理模型

Gemini 3.1 Pro Preview is Google’s frontier reasoning model, delivering enhanced software engineering performance, impro… 输入 $2.00/M · 输出 $12.00/M

— 暂无评测数据

输入 $2.00/M 输出 $12.00/M

Anthropic: Claude Sonnet 4.6

2026-02-17 Anthropic

Sonnet 4.6 is Anthropic's most capable Sonnet-class model yet, with frontier performance across coding, agents, and prof… 输入 $3.00/M · 输出 $15.00/M

综合 95.3 基础 96 进阶 95 困难 95 L 95

输入 $3.00/M 输出 $15.00/M

GLM-5：从氛围编程到智能体工程

2026-02-17 z.ai

论文大模型 GLM 开源

智谱AI发布GLM-5系列技术报告，标志着模型从单纯的文本生成向具备复杂推理与自主行动能力的智能体进化。GLM-5在架构上实现了原生多模态理解与生成的深度融合，重点强化了长文本处理、复杂逻辑推理及工具调用能力。报告详细介绍了从“Vibe Coding”向“Agentic Engineering”转型的技术路径，通过创新的训练范式提升了模型在真实世界任务中的可靠性。作为国产大模型的代表作，GLM-5在多个权威基准测试中达到国际领先水平，为构建通用人工智能智能体提供了重要技术支撑。

综合 92.4 基础 95 进阶 92 困难 89

输入 $0.55/M 输出 $2.48/M

Qwen: Qwen3.5 Plus 2026-02-15

2026-02-16 Qwen

The Qwen3.5 native vision-language series Plus models are built on a hybrid architecture that integrates linear attentio… 输入 $0.40/M · 输出 $2.40/M

综合 95.1 基础 96 进阶 95 困难 95 L 95

输入 $0.12/M 输出 $0.69/M

MiniMax: MiniMax M2.5

2026-02-12 MiniMax

MiniMax-M2.5 is a SOTA large language model designed for real-world productivity. Trained in a diverse range of complex … 输入 $0.30/M · 输出 $1.10/M

综合 75.9 基础 78 进阶 76 困难 76 L 87 W 63

输入 $0.30/M 输出 $1.10/M

2026-02-11 Z.ai

多模态代码推理模型

GLM-5 is Z.ai’s flagship open-source foundation model engineered for complex systems design and long-horizon agent workf… 输入 $0.30/M · 输出 $2.55/M

综合 92.4 基础 95 进阶 92 困难 89 L 92

输入 $0.55/M 输出 $2.48/M

Qwen: Qwen3 Max Thinking 待评测

2026-02-09 Qwen

Qwen3-Max-Thinking is the flagship reasoning model in the Qwen3 series, designed for high-stakes cognitive tasks that re… 输入 $1.20/M · 输出 $6.00/M

— 暂无评测数据

输入 $1.20/M 输出 $6.00/M

Anthropic: Claude Opus 4.6 待评测

2026-02-04 Anthropic

Opus 4.6 is Anthropic’s strongest model for coding and long-running professional tasks. It is built for agents that oper… 输入 $5.00/M · 输出 $25.00/M

— 暂无评测数据

输入 $5.00/M 输出 $25.00/M

1 月

MoonshotAI: Kimi K2.5 待评测

2026-01-27 Moonshotai

多模态推理模型

Kimi K2.5 is Moonshot AI's native multimodal model, delivering state-of-the-art visual coding capability and a self-dire… 输入 $0.23/M · 输出 $3.00/M

— 暂无评测数据

输入 $0.23/M 输出 $3.00/M

Z.ai: GLM 4.7 Flash 待评测

2026-01-19 Z.ai

As a 30B-class SOTA model, GLM-4.7-Flash offers a new option that balances performance and efficiency. It is further opt… 输入 $0.06/M · 输出 $0.40/M

— 暂无评测数据

输入 $0.06/M 输出 $0.40/M

2025

12 月

Z.ai: GLM 4.7 待评测

2025-12-22 Z-ai

GLM-4.7 is Z.ai’s latest flagship model, featuring upgrades in two key areas: enhanced programming capabilities and more stable multi-step reasoning/execution. It demonstrates significant improvements…

— 暂无评测数据

输入 $0.38/M 输出 $1.70/M

Google: Gemini 3 Flash Preview

2025-12-17 Google

多模态闭源

Gemini 3 Flash Preview is a high speed, high value thinking model designed for agentic workflows, multi turn chat, and coding assistance. It delivers near Pro level reasoning and tool use performance…

综合 82.2 基础 84 进阶 82 困难 82 L 88 W 67

输入 $0.50/M 输出 $3.00/M

Xiaomi: MiMo-V2-Flash 待评测

2025-12-14 Xiaomi

MiMo-V2-Flash is an open-source foundation language model developed by Xiaomi. It is a Mixture-of-Experts model with 309B total parameters and 15B active parameters, adopting hybrid attention architec…

— 暂无评测数据

输入 $0.09/M 输出 $0.29/M

OpenAI: GPT-5.2

2025-12-10 OpenAI

多模态闭源

GPT-5.2 is the latest frontier-grade model in the GPT-5 series, offering stronger agentic and long context perfomance compared to GPT-5.1. It uses adaptive reasoning to allocate computation dynamicall…

综合 96.1 基础 97 进阶 96 困难 96 L 96

输入 $1.75/M 输出 $14.00/M

2025 年 10 月 — 11 月 · 无发布记录

9 月

Anthropic: Claude Sonnet 4.5 待评测

2025-09-29 Anthropic

多模态闭源

Claude Sonnet 4.5 is Anthropic’s most advanced Sonnet model to date, optimized for real-world agents and coding workflows. It delivers state-of-the-art performance on coding benchmarks such as SWE-ben…

— 暂无评测数据

输入 $3.00/M 输出 $15.00/M

8 月

7 月

Google: Gemini 2.5 Flash Lite

2025-07-22 Google

多模态闭源

Gemini 2.5 Flash-Lite is a lightweight reasoning model in the Gemini 2.5 family, optimized for ultra-low latency and cost efficiency. It offers improved throughput, faster token generation, and better…

综合 66.4 基础 71 进阶 66 困难 64 L 70 W 57

输入 $0.10/M 输出 $0.40/M

— 已到达最早记录 —