跳到主内容
@wquguru
精选88Latent Space(RSS)模型发布/更新多源精选 ×10

OpenAI发布GPT-6.2 Astra与Sol,成本大幅降低但评价两极

[AINews] GPT-6.2 Astra, the #2 Frontier Model is better and 20%-600% cheaper, but has mixed reviews; 6.2 Sol likely main workhorse

原文
发到 X
推荐理由

GPT-6.2是全新旗舰迭代,不仅涉及核心模型发布,更带来了显著的成本结构变化与能力跃迁,对行业竞争格局有重大影响,值得重点关注。

OpenAI shipped GPT-6.2 Astra and Sol today, two days before Anthropic’s expected first trading day. In Artificial Analysis’ tests, Sol ties September’s Astra at about 15% of the cost per task. New Astra scores 62 on the Intelligence Index, just behind Fable 5.2 at 65. Net net, Anthropic still owns the world’s best model, but OpenAI owns the world’s most efficient models in terms of intelligence per $.

OpenAI 今日发布了 GPT-6.2 Astra 和 Sol,比 Anthropic 预计的首个交易日早两天。在 Artificial Analysis 的测试中,Sol 的任务成本约为 Astra(9月版本)的 15%,两者持平。新版 Astra 在 Intelligence Index 上得分为 62,略低于 Fable 5.2 的 65 分。总体而言,Anthropic 仍拥有全球最佳模型,但 OpenAI 拥有按每美元智能度计算效率最高的全球模型。

What’s more unusual about this launch is the STRONGLY negative influencer feedback…

此次发布更不寻常之处在于影响者给出了强烈负面的反馈……

which prompted a defensive response…

这引发了一种防御性的回应……

View post on X ↗

在 X 上查看帖子 ↗

We think that some of this is overreacting and normal influencer tendency to either have everything be 🤯🤯🤯 or the literal WORST.MODEL.EVER, BUT also that some of the criticisms are valid, and the RL artefacts are noticeable and downright bizarre (don’t even get us started about the terrible launch video). However, we still use it to automate our work and that’s all that matters in the end.

我们认为其中一些反应属于过度反应,这是影响者的正常倾向,要么觉得一切令人惊叹🤯🤯🤯,要么认为是史上最烂.MODEL.EVER,但同时也认为部分批评是合理的,强化学习(RL)产生的伪影明显且相当离奇(甚至别提那糟糕的发布视频了)。不过,我们仍然使用它来自动化我们的工作,这才是最终重要的事。

Great ship, but Anthropic’s IPO is unaffected and that will be probably the headline of the week/year/decade (until OpenAI IPO’s….)

出色的发布,但 Anthropic 的 IPO 不受影响,这可能会成为本周/今年/本十年(直到 OpenAI IPO……)的头版新闻。

Today’s AI News is brought to you by SerpApi!

今天的 AI 新闻由 SerpApi 赞助播出!

Retrieve public web pages by URL as LLM-ready Markdown or structured JSON. Map sites and run bounded crawls with one API. Pair with 100+ SerpApi search APIs for search-to-content workflows.

通过 URL 检索公共网页,转换为适合大语言模型的 Markdown 或结构化 JSON。使用一个 API 映射网站并运行有界爬取。结合 100 多个 SerpApi 搜索 API,实现从搜索到内容的流程。

Check them out to support LS, and contact [email protected] for feedback!

请支持 LS,如有反馈请联系 [email protected]!

AI Twitter Recap

AI Twitter 回顾

GPT-6.2 Astra and Sol

GPT-6.2 Astra 和 Sol

  • Availability. Today’s models are rolling out to ChatGPT Work and Codex for Plus, Pro, Enterprise and Edu, and to developers through the API as gpt-6.2-astra and gpt-6.2-sol. OpenAI
  • API prices per million tokens. Astra is $8 input, $0.40 cached input and $40 output. Sol keeps its predecessor’s pricing at $2, $0.10 and $10. Both have 1M context and a 128K output limit. Vals · Ultrafast pricing and rollout
  • Astra for Law. The Enterprise configuration adds a legal search index. Artificial Analysis measured 20.8% on Harvey LAB-AA, versus 15.0% for base Astra. Artificial Analysis
  • Preparedness. OpenAI rated Astra High for AI Self-Improvement, a first for one of its externally released models. On a separate measure, Vals’ RSI-Index rose from 27.07% to 29.01%, with 50% denoting its human reference. System card · Vals
  • FrontierMath Erdős. Epoch verified seven solutions in Lean out of 68 problems. The previous best was three. Epoch
  • 可用性。今日发布的模型正在向 ChatGPT Work 和 Codex 的 Plus、Pro、Enterprise 和 Edu 用户推出,并通过 API 向开发者提供 gpt-6.2-astra 和 gpt-6.2-sol。OpenAI
  • API 每百万代币价格。Astra 为输入 $8、缓存输入 $0.40、输出 $40。Sol 保持其前身定价:$2、$0.10 和 $10。两者均具有 1M 上下文窗口和 128K 输出限制。Vals · Ultrafast 定价和发布
  • Astra for Law。企业配置增加了法律搜索索引。Artificial Analysis 在 Harvey LAB-AA 上的测评分为 20.8%,而基础版 Astra 为 15.0%。Artificial Analysis
  • 准备情况。OpenAI 将 Astra 在 AI 自我改进方面的评级评为高,这是其外部发布模型中的首次。在另一项指标上,Vals 的 RSI-Index 从 27.07% 上升至 29.01%,其中 50% 代表人类参考水平。系统卡片 · Vals
  • FrontierMath Erdős。Epoch 在 68 个问题中验证了 7 个 Lean 解决方案。此前的最佳成绩是三个。Epoch

What the benchmarks measure

基准测试衡量的是什么

  • Agents’ Last Exam 1.1. OpenAI reports Astra at xhigh gained 5.2 points over GPT-6 Astra at max, using less than a third of the cost per task. It reports a 5.5-point lead over Fable 5.2 at about 5% of the cost on that test. On GDPval v2, it says Astra used 22% fewer output tokens than its predecessor. OpenAI’s results
  • Long engineering runs. Mercor’s OpenWorldEngineer puts Astra at 67.2%, Fable 5.2 at 54.6%, and GPT-6 Astra at 44.2%. On Vals’ multi-turn Vibe Code Bench, Astra was first and Sol was third. Mercor · Vals
  • 3D and robotics. Astra led AA-3D with 1,318, ahead of Fable at 1,271 and Sol at 1,259. On Artificial Analysis’s robotics preview, Astra scored 79 and Sol 78, ahead of Fable at 75. Artificial Analysis
  • Research agents. On Perplexity’s WANDR, Astra scored 0.783 at about $12.54 per task. Sol reached 0.746 at about $2.65. Perplexity
  • 智能体最新评测 1.1。OpenAI 报告称,Astra 在 xhigh 设置下比 GPT-6 Astra 高出 5.2 分,且每项任务成本不到后者的三分之一。它在该测试中比 Fable 5.2 领先 5.5 分,成本约为后者的 5%。在 GDPval v2 上,它表示 Astra 的输出 token 数比其前身减少了 22%。OpenAI 的结果
  • 长期工程运行。Mercor 的 OpenWorldEngineer 将 Astra 排在 67.2%,Fable 5.2 为 54.6%,GPT-6 Astra 为 44.2%。在 Vals 的多轮 Vibe Code Bench 上,Astra 排名第一,Sol 排名第三。Mercor · Vals
  • 3D 与机器人技术。Astra 以 1,318 分领跑 AA-3D,高于 Fable 的 1,271 分和 Sol 的 1,259 分。在 Artificial Analysis 的机器人预览中,Astra 得分为 79,Sol 为 78,高于 Fable 的 75 分。Artificial Analysis
  • 研究智能体。在 Perplexity 的 WANDR 上,Astra 得分为 0.783,每项任务成本约 12.54 美元。Sol 得分为 0.746,每项任务成本约 2.65 美元。Perplexity

Early testers disagreed

早期测试者意见不一

  • Matt Shumer called the models “essentially unusable” after weeks of testing and said his Opus and Fable agents rejected much of Astra’s subagent work. Review · Workflow details
  • Pietro Schirano attributed many of his early issues to the harness. Arena’s Peter Gostev tested Astra against Fable on visual coding and singled out Astra’s SVG peacock. Schirano · Gostev
  • The demos. Gabriel Chua built a walkable Singapore using more than 100 Codex threads. Simon Willison ran the same three.js mech prompt through several models, including Astra, Sol, Carbon and Fable. Singapore build · Same-prompt comparison
  • Matt Shumer 在数周测试后称这些模型“基本上无法使用”,并表示他的 Opus 和 Fable 智能体拒绝了 Astra 大量子智能体的工作成果。评论 · 工作流详情
  • Pietro Schirano 将许多早期问题归因于测试框架。Arena 的 Peter Gostev 对 Astra 和 Fable 进行了视觉编码测试,并特别指出了 Astra 生成的 SVG 孔雀。Schirano · Gostev
  • 演示环节。Gabriel Chua 使用超过 100 个 Codex 线程构建了一个可交互的新加坡。Simon Willison 将相同的 three.js 机甲提示词输入多个模型,包括 Astra、Sol、Carbon 和 Fable。新加坡构建 · 同提示词对比

Microsoft Ignite and Gemini

Microsoft Ignite 与 Gemini

  • Microsoft ships its own coding models. MAI Code 2 is for multi-step coding, and Code 2 Flash is for quicker edits. Both are available in GitHub Copilot’s model picker, Copilot Fusion and Foundry. Microsoft also announced a multimodal upgrade to its Decision model. Microsoft
  • Copilot Fusion. Microsoft reports 71.2% on Terminal-Bench 4.1 with its multi-model harness, at 88% lower cost than Claude Fable 5.2 in that comparison. Jay Parikh
  • Image editing. Artificial Analysis ranks MAI Image 2.7 first on its editing benchmark, at $34.82 per 1,000 images. Artificial Analysis
  • Gemini Tasks. Hydrogen now powers free Tasks in six countries, including tasks that need a phone call or a text. Google also showed a new Antigravity robotics plugin with MuJoCo simulation. Sundar Pichai · Google DeepMind
  • 微软推出自有编码模型。MAI Code 2 用于多步编码,Code 2 Flash 用于快速编辑。两者均已在 GitHub Copilot 的模型选择器、Copilot Fusion 和 Foundry 中提供。微软还宣布对其 Decision 模型进行多模态升级。Microsoft
  • Copilot Fusion。微软报告称,其多模型测试框架在 Terminal-Bench 4.1 上得分 71.2%,在与 Claude Fable 5.2 的对比中成本降低 88%。Jay Parikh
  • 图像编辑。Artificial Analysis 在其编辑基准测试中将 MAI Image 2.7 列为第一,每 1,000 张图像成本为 34.82 美元。Artificial Analysis
  • Gemini Tasks。Hydrogen 现在在六个国家提供免费 Tasks 服务,包括需要电话或短信的任务。Google 还展示了一个带有 MuJoCo 模拟的新 Antigravity 机器人插件。Sundar Pichai · Google DeepMind

Agent harnesses and inference

智能体测试框架与推理

Hugging Face’s Pi adapter. A small adapter trained inside pi lifted Qwen3.8-27B from 15 to 32 solves on the 51 hardest SWE-bench Pro tasks. Across that subset it made 55% fewer tool calls per task. Model card and paired results

Hugging Face 的 Pi 适配器。在 pi 内部训练的小型适配器将 Qwen3.8-27B 在 SWE-bench Pro 最难的 51 个任务上的解决数从 15 提升至 32。在该子集上,每个任务的工具调用次数减少了 55%。模型卡片和配对结果

  • Harness costs vary sharply. Baseten measured up to a 4.4× spread in cost per passing task across its benchmark of 22 harnesses and five open models. Baseten
  • The harness may specialize to its model. Google’s SIM keeps the rationale and outcome of prior harness edits. A harness it evolved on Opus scored 68.41 on Gemini, versus 73.89 for the starting harness. Research summary
  • More multi-agent products. Cognition launched persistent Devin Bots. Perplexity reported that its mixed-model teams beat four instances of the lead model on Terminal-Bench and WANDR. Replit reported diminishing returns from 16 to 32 agents on NumPy tasks. Cognition · Perplexity · Replit
  • Coordination dominated a 1,000-agent run. Hugging Face says only 40–60 agents were calling the model at any moment. They read roughly 130 tokens for every token they wrote, with merges and tests contributing to the bottleneck. Thomas Wolf
  • A small fraction of cache misses can dominate the bill. SemiAnalysis reports that 1.8% of uncached requests accounted for roughly 15% of spend in its sample. SemiAnalysis
  • Marin and K2 Horizon opened up their training runs. Marin released 57 intermediate Mochi checkpoints from its 535B-A23B run. The K2 report compares four model sizes trained on the same 22T tokens and describes a decoding adapter that the team says reaches 2.2× speedup. Marin · K2 Horizon
  • Zyphra’s long-context attention. The team reports 9–60× speedups at 10M tokens after converting several models to its subquadratic attention, with Qwen losing under two points on RULER at 1M. Zyphra
  • Speculative decoding. vLLM reports up to 30% higher acceptance length from a sliding-window draft for Kimi K3.1 Pro. NVIDIA also has a 4B draft for Qwen3.8-27B. vLLM · Draft-model research
  • Harness 成本差异显著。Baseten 在其包含 22 个 harness 和五个开源模型的基准测试中,测得通过任务的单位成本最高相差 4.4 倍。Baseten
  • Harness 可能会针对其特定模型进行优化。Google 的 SIM 保留了先前 harness 编辑的依据和结果。一个在 Opus 上进化而来的 harness 在 Gemini 上得分 68.41,而初始 harness 得分为 73.89。研究摘要
  • 更多多智能体产品。Cognition 推出了持久化的 Devin Bots。Perplexity 报告称,其混合模型团队在 Terminal-Bench 和 WANDR 上击败了主模型的四个实例。Replit 报告称在 NumPy 任务上,智能体数量从 16 增加到 32 时出现了收益递减现象。Cognition · Perplexity · Replit
  • 协调性主导了 1,000 个智能体的运行。Hugging Face 表示,任何时刻只有 40–60 个智能体在调用模型。它们每写入一个 token 大约读取 130 个 token,合并(merges)和测试(tests)导致了瓶颈。Thomas Wolf
  • 一小部分缓存未命中可能主导账单。SemiAnalysis 报告称,1.8% 的未缓存请求占其样本支出的约 15%。SemiAnalysis
  • Marin 和 K2 Horizon 公开了其训练过程。Marin 发布了其 535B-A23B 运行中的 57 个中间 Mochi 检查点。K2 的报告比较了在相同 22T token 上训练的四种模型大小,并描述了一种解码适配器,据称该适配器可实现 2.2 倍的加速。Marin · K2 Horizon
  • Zyphra 的长上下文注意力机制。该团队报告称,在将多个模型转换为其亚二次方注意力机制后,在 10M token 长度下实现了 9–60 倍的加速,Qwen 在 1M token 下的 RULER 测试中损失不到两个百分点。Zyphra
  • 推测解码。vLLM 报告称,对于 Kimi K3.1 Pro,使用滑动窗口草稿(sliding-window draft)可使接受长度提高高达 30%。NVIDIA 也为 Qwen3.8-27B 提供了 4B 大小的草稿模型。vLLM · Draft-model research

Industry and policy

行业与政策

  • NVIDIA’s first earnings after the CEO transition. Under interim CEO Colette Kress, NVIDIA posted $115.8B in quarterly revenue, up 103% year over year, and said Vera Rubin NVL72 had shipped. The board said it did not intend to reinstate Jensen Huang. NVIDIA · AP
  • Isomorphic Labs raises $5B. Greenoaks and Sequoia co-led the Series C. The company says it will scale its drug-design platform and develop its therapeutic pipeline. Isomorphic Labs
  • Genesis AI raises $500M. The Series A values the robotics company at $3.5B post-money. It plans to use the funding to deploy its Eno robot and expand training and data collection. Zhou Xian
  • CEO 交接后的 NVIDIA 首次财报。在临时 CEO Colette Kress 的带领下,NVIDIA 季度营收达到 1158 亿美元,同比增长 103%,并表示 Vera Rubin NVL72 已发货。董事会表示无意恢复 Jensen Huang 的职位。NVIDIA · AP
  • Isomorphic Labs 融资 50 亿美元。Greenoaks 和 Sequoia 联合领投 C 轮融资。该公司表示将扩大其药物设计平台规模并开发治疗管线。Isomorphic Labs
  • Genesis AI 融资5亿美元。A轮融资后,这家机器人公司的估值达到35亿美元。该公司计划利用这笔资金部署其Eno机器人,并扩大培训和数据收集范围。周贤

AI Reddit Recap

AI Reddit 回顾

Local models and hardware

本地模型与硬件

  • A 9B model with layers that run twice. SenseNova-Loop-9B runs 24 of its 32 layers a second time. The model’s authors report GAIA rising from 37.86 in the preview to 69.90 in the release. Local users asked how much memory remains for the KV cache at long context. Weights · Discussion
  • 一个具有双层结构的9B模型。SenseNova-Loop-9B 会将其32层中的24层再运行一次。该模型的作者报告称,GAIA基准测试分数从预览版的37.86提升至发布版的69.90。本地用户询问在长上下文场景下,KV缓存还剩多少内存。权重 · 讨论

更进一步:量化金融体系

看懂新闻只是起点——沿量化金融路径,把它变成能交付的工程能力

进入量化体系 →

关联讨论

同一事件的更多信源

相似阅读

关联信息,但可能不是同一事件