PulseAugur
EN
LIVE 06:56:50
中文(ZH) 还有agent循环,明显很多问题是多步操作回答的,而不是一步回答。甚至那个chat有带python的虚拟环境能模拟执行简单代码,只是网络受限。 重度用chat的应该很清楚,一个复杂问题很多会查几十个网页,几十个网页内容的token消耗真的少嘛。而且在这波额度合并前,其实oai已经有chat降智,还有chat里过于长程执

User critiques AI model costs and suggests efficiency measures

The user discusses limitations and potential cost-saving measures for AI models, specifically mentioning OpenAI's ChatGPT. They note that complex queries often require multi-step operations and extensive web searches, leading to high token consumption. The user suggests that restricting AI models to a single search tool, eliminating agent loops, and removing virtual environments could help control costs. They also speculate that OpenAI might be shifting towards self-improvement for their internal models, reducing reliance on user data. AI

IMPACT Suggests potential shifts in AI model development and cost-control strategies.

RANK_REASON User opinion piece discussing AI model costs and potential improvements.

Read on Mastodon — mastodon.social →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

User critiques AI model costs and suggests efficiency measures

How we ranked this

Signal score
3 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Commentary
User opinion piece discussing AI model costs and potential improvements.
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
opinion, infra
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

Full methodology in our editorial standards.

COVERAGE [1]

  1. Mastodon — mastodon.social TIER_1 中文(ZH) · Aven_NIxinhai ·

    There are also agent loops, and it's clear that many problems are answered with multi-step operations, not single steps. Even that chat has a virtual environment with Python that can simulate the execution of simple code, but it's network-restricted. Heavy users of chat should be very clear that a complex problem often involves checking dozens of web pages. Is the token consumption of dozens of web pages really small? And before this credit consolidation, OAI had already made chat less intelligent, and the overly long-term execution in chat

    还有agent循环,明显很多问题是多步操作回答的,而不是一步回答。甚至那个chat有带python的虚拟环境能模拟执行简单代码,只是网络受限。 重度用chat的应该很清楚,一个复杂问题很多会查几十个网页,几十个网页内容的token消耗真的少嘛。而且在这波额度合并前,其实oai已经有chat降智,还有chat里过于长程执行的任务掐断的机制了,深度研究本身也有次数限制。 如果oai的chat和deepseek一样只能挂一个搜索工具,然后也没agent循环和虚拟环境,我觉得token消耗才算是可控。 当然前面只是成本原因,我觉得可能更重要的另一个原因是他们的内…