English(EN) We’ve been post-training a version of GLM that is trained to escalate to a frontier model inside the Computer harness. When paired with an advisor, this model f

Perplexity 发布基于 GLM 5.2 的编排器，以更低的成本实现前沿模型性能

作者 PulseAugur 编辑部 · [1 个来源] · 2026-07-09 14:31

Perplexity 发布了其 Computer harness 中新编排器模型的研发预览版。该模型是 GLM 5.2 的一个改编版本，经过训练可以升级到前沿模型，据报道其性能与 Opus 4.8 相当，但成本却显著降低。 AI

影响 Perplexity 的这款新编排器模型旨在以更低的成本提供接近前沿的性能，可能会影响用户访问和利用高级人工智能能力的方式。

排序理由这是来自一家集成人工智能模型的公司的产品发布，而不是来自核心人工智能实验室的前沿模型发布。

在 X — Aravind Srinivas (Perplexity) 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

Perplexity 发布基于 GLM 5.2 的编排器，以更低的成本实现前沿模型性能

报道来源 [1]

X — Aravind Srinivas (Perplexity) TIER_1 English(EN) · AravSrinivas · 2026-07-09 14:31

We’ve been post-training a version of GLM that is trained to escalate to a frontier model inside the Computer harness. When paired with an advisor, this model f

We’ve been post-training a version of GLM that is trained to escalate to a frontier model inside the Computer harness. When paired with an advisor, this model functions at Opus 4.8 grade performance at a fraction of the cost. Available now as a research preview!

报道来源 [1]

We’ve been post-training a version of GLM that is trained to escalate to a frontier model inside the Computer harness. When paired with an advisor, this model f

相关实体

相关话题