OpenAI Repeatedly Altered GPT-6 Astra Benchmark Data, Some Scores Changed Significantly
Bubble signal: 0 · NeutralRelevance 30ModelsUS
Source: IT Home · Published 2026-09-06
Summary
According to Fortune, OpenAI has repeatedly modified benchmark data in its GPT-6 Astra announcement after release, including lowering Astra's hallucination rate from 4.2% to 2% before reverting. These changes followed technical issues and publication delays, raising questions about benchmark reliability and transparency in AI marketing.
Bubble analysis
This news does not directly involve investment amounts or valuations, but it reflects intense competition and marketing pressure among AI companies, potentially leading to overhyped claims that inflate market expectations about AI capabilities, indirectly affecting the investment bubble.
#benchmark#openai#gpt-6
Read original ↗Related signals
linked by shared tags
- OpenAI Dot VM Specs Leaked: AMD EPYC 9V74 CPU + 9.73GB RAM, 9435 Multi-Core Score on Geekbench 7#openai#gpt-6
- OpenAI Launches Round-the-Clock Autonomous Agent Dots and GPT-6.1 Sol Model#openai#gpt-6
- OpenAI's New Essay Pulls Superintelligence Back to the Execution Layer: Even a Billion Einsteins Need Someone to Mine#openai
- Racing to List Before Thanksgiving: Anthropic Said to Seek IPO as Early as Mid-November#openai
- OpenAI's new agent is a shot at Meta — but can it compete with free?#openai