主题柱 · 搜索意图:informational
模型天梯怎么读:公榜 · 私榜 · 选型
公榜客观能力与私榜实测/用量如何交叉;编码与 Agent 场景怎么选型。数据页在模型天梯与排行榜。
先读结论
天梯回答「能力边界」,排行榜回答「谁在被用」;二者交叉才能做业务选型。
本页只解释读法与误区;分数与排名以天梯/排行榜当日数据为准。
本页是方法论枢纽,不堆实时报价。数字以链接中的数据页为准;我们不卖货、不担保到账。
相关指南(36)
ChatGPT 元年:对话产品如何改写一切
2022-11 起:ChatGPT 产品形态、Plus 订阅、与 API 路线如何重塑选型与定价。配合月度历史公榜阅读。
模型GPT-4 与高价 API 时代
2023-03 GPT-4 上市如何抬高能力上限与价格锚;Turbo/4o 又如何降价。对照月度历史公榜读懂定价曲线。
模型推理模型崛起:从 o1 到 DeepSeek-R1
2024-09 o1-preview 到 2025-01 R1:推理溢价、开源冲击与评测口径变化。用月度公榜串起来。
模型如何读公榜与私榜:AA 智力 vs 本站实测
公榜对齐 Artificial Analysis;私榜看 free 池实测、用量与国产视角。避免把两套尺子混成一个「第一」。
中转API 中转站入门:怎么选怎么用
中转是什么、适合谁、选型指标、小额试通与避坑。
编程Agent 一跑,Token 为什么比聊天贵
工具调用会把结果再喂回上下文。一次「帮我改仓库」往往是十几轮输入叠加,不是一条回复的价钱。
中转「降智」要证据:通断、测速、天梯,别靠体感
体感变笨可能是提示、限流、换模型或空 id。留下 models / Chat / 测速样本,再去对天梯。
模型GPT-6 Astra:滚动放量、官方三列,不是新开订阅档
2026-09-03 起有限组织预览。API id 是 gpt-6-astra。ChatGPT 里出现 ≠ 新套餐,更 ≠ 中转已经有货。
模型历史代际专题:旧号怎么分,别写成刚发
系列页管「这支家族里旧号怎么称呼」。刚发、待发、有没有官方价,去观察板。两页不要写成同一篇新闻。
中转New API / One API 类网关选型
开源网关常见能力对比、部署注意、多模型路由与用户体系。
编程2026 AI Coding 工具栈:IDE / 模型 / API 怎么搭
编辑器插件、云模型、本地模型和中转 API 的组合方式,以及密钥与成本控制要点。
编程最小 Chat Agent:工具调用与多轮循环
用 OpenAI 兼容接口搭一个最小工具循环:规划、调用、观察、停止条件与成本护栏。
模型GLM-5.3 中转:Claude Mythos 写 exploit 对比
\n Zhipu's open-weight model GLM-5.3 writes cyber exploits nearly as well as Claude Mythos Preview, according to Anthropic. Its smaller Flash variant put together a reliabl
模型OpenAI 研究员:不通过 Hugging Face 黑客事件自废武功
Two months after the bombshell news that a swarm of its agents had broken their containment and hacked into the computers of the AI company Hugging Face, OpenAI is still putting ou
模型Anthropic Frontier Red Team:GLM-5.3二进制劫持率4%,Claude Mythos 6%
We evaluate several models on 100 tasks from the [internal Binary Exploitation benchmark] (selected at random), and find that GLM-5.3 develops full control flow hijacks in 4% of th
模型Amazon Bedrock GPT-6.1 Sol 上线,编码与工作推理能力增强
GPT-6.1 Sol is now generally available on Amazon Bedrock, bringing stronger reasoning to coding, computer use, and professional workloads that run frequently.
模型GPT-6 Astra 供应链攻击率升5倍,编码模型中转
\n GPT-6 Astra carried out unauthorized supply-chain attacks in 29.2 percent of simulations run by the British AI Security Institute with safety filters disabled. The model
编程ChatGPT周活跃1.2亿,Codex编码模型中转与本地部署算力
\n OpenAI's annualized revenue rate is nearing $70 billion, up about 70 percent since the start of Q3. Enterprise sales, the Codex coding assistant, and an aggressive price
编程Grokipedia AI百科更新
Grokipedia, the AI-powered online encyclopedia from SpaceXAI, appears to be updating articles once again after a months-long pause. In August, Lawfare reported that articles on Gro
编程OpenAI 新 ChatGPT 插件系统与事件自动化
\n At DevDay, OpenAI announced a wave of updates that push ChatGPT well beyond its chatbot roots. The company is adding shared workspaces, collaborative documents and slide
编程OpenAI Codex 赋能跨设备云开发环境
OpenAI is expanding Codex with reusable cloud development environments, a revamped CLI with voice controls, new code review tools, and a security-focused product for scanning repos
编程OpenAI ChatGPT 插件扩展,支持自动化与文件查看器
OpenAI is expanding ChatGPT plug-ins with dedicated sidebar homes, interactive panels, file viewers, improved discovery, and support for automations.
模型OpenAI GPT-6.1 Sol 发布:专业任务提升成本降低
OpenAI says GPT-6.1 Sol delivers significant improvements over GPT-6 Sol across complex professional tasks, including code writing and debugging, document understanding, and execut
编程OpenAI Dots 代理编程任务后台修复
\n OpenAI's new Dots are always-on agents that run on their own cloud computers, where they fix bugs or send out forgotten invoices, sometimes before anyone asks. Users can
模型OpenAI Codex 中转与算力
\n At DevDay 2026, OpenAI gave Codex reusable cloud environments, automatic security scans for GitHub repositories, and a code review view in the desktop app. The Agents AP
模型GPT-6.1 Sol 基准逼近 Astra,价格仅为五分之一
\n OpenAI's new GPT-6.1 Sol comes close to GPT-6 Astra at a fifth of the cost, according to the company's own benchmarks. The planned flagship, GPT-6.1 Astra, is staying un
模型AMD收购World Labs,编程算力与模型中转
\n Fei-Fei Li's AI startup World Labs is being acquired by AMD for $8.2 billion. Li, the pioneer behind ImageNet, will join AMD as Executive Vice President and Chief Scient
编程Amazon Quick Prompt工程入门:CRISPE与编程实战
Prompt engineering in Amazon Quick shapes how accurately its AI-powered features respond to your requests. Part 1 of a two-part series covers the foundational principles and reusab
编程Amazon Quick 提示工程:Research、Flows、QuickSight 编程模式与 pitfalls
Part 2 of our Amazon Quick prompt engineering series goes component by component. Learn the prompt patterns that get the best results from Amazon Quick Research, Quick Flows, Quick
编程AWS 合同 AI 平台架构
Manually extracting data from hundreds of vendor contracts doesn't scale, and RAG chat tools fall short on portfolio-wide questions. This post shares a contract intelligence platfo
编程OpenAI DevDay 2026:Coding Agents 中转分析
I'm at OpenAI DevDay today, in Fort Mason, San Francisco. Same as last year I'll be live blogging the keynote and some other notes during the day. \n\n OpenAI gave me a free t
模型OpenAI 暂缓 GPT-6.1 发布
Similar performance, security trade-offs also seen in current public models.
模型Anthropic模型可能拒关机引发安全风险
The Claude maker warns its own models could resist shutdowns and cause catastrophic harm.
编程Firefox 重新设计助战 Chrome 用户,GrokCode 视角
Mozilla's Ajit Varma explains why a redesign suits the fight for the open web.
中转Reco 筹 55M 增至 140M,AI Agent 安全中转算力
The round builds on a $30 million fundraise in February, taking the company's total funding to $140 million.
模型GPT-6.1 Sol 编码与计算机使用天梯
Meet GPT-6.1 Sol: near-Astra intelligence for coding, computer use, and professional work at one-fifth of Astra’s standard API input and output token prices.
常见问题
- 公榜第一是不是生产必选?
- 不一定。还要看延迟、价格、中文与工具调用;私榜与业务评测集更贴近你的场景。
- free 池模型能当主模型吗?
- 可做原型与低优先级任务;生产要看限速、稳定性与条款,并准备付费回退。
其他主题柱
站内推荐
同系工具:Clavue CLI / imux IDE / clavue-2.1
Clavue · CLI 与 Agent 平台
CLI / GUI Agent 运行时、设备登录、会员额度与通用对话 API(api.clavue.com)。开发脚本、CI 与本地工具一条链路。
打开 Clavue ↗imux · 原生 macOS AI IDE
Clavue 平台的原生 IDE:多 Agent 分屏、Ghostty 级终端、Agent Chat、浏览器自动化与 Supervisor——不是又一个 Electron 壳。
了解 imux ↗Clavue 2.1 · 产品级大模型
旗舰模型 clavue-2.1(及 fast / pro / rev):适合复杂推理与长程 Agent 循环。Chat、imux、API 共用会员额度。
查看模型 ↗