GPT-6 Sol · User-flagged factuality, high-difficulty prompts · Anonymized ChatGPT conversations previously flagged by users for factual errors; exact set version and sample size unspecified · 成绩与配置
GPT-6 Sol 的 User-flagged factuality, high-difficulty prompts · Anonymized ChatGPT conversations previously flagged by users for factual errors; exact set version and sample size unspecified 记录。本次发布有 5 条成绩记录,含 5 条精确值(可含历史记录),附 5 条来源定位。
GPT-6 Sol · User-flagged factuality, high-difficulty prompts 成绩与配置
指标与成绩
配置与评测条件
出处
Responses with at least one factual error11.4%原文 11.4% · 单位 %
公布 未确认
GPT-6 Sol low
设置
GPT-6 Sol; reasoning effort low as directly labeled in the official chart.
未知项
未另记
工具
工具条件未记录
推理预算
low
框架
未记录
协议
High-difficulty anonymized conversations previously flagged by users for factual errors; an answer counts as an error when it contains at least one factual error; scores are not length-adjusted.
条件
GPT-6 Sol; reasoning effort low. The selected prompts are not representative of typical use.
Responses with at least one factual error6.9%原文 6.9% · 单位 %
公布 未确认
GPT-6 Sol medium
设置
GPT-6 Sol; reasoning effort medium as directly labeled in the official chart.
未知项
未另记
工具
工具条件未记录
推理预算
medium
框架
未记录
协议
High-difficulty anonymized conversations previously flagged by users for factual errors; an answer counts as an error when it contains at least one factual error; scores are not length-adjusted.
条件
GPT-6 Sol; reasoning effort medium. The selected prompts are not representative of typical use.
Responses with at least one factual error5.1%原文 5.1% · 单位 %
公布 未确认
GPT-6 Sol high
设置
GPT-6 Sol; reasoning effort high as directly labeled in the official chart.
未知项
未另记
工具
工具条件未记录
推理预算
high
框架
未记录
协议
High-difficulty anonymized conversations previously flagged by users for factual errors; an answer counts as an error when it contains at least one factual error; scores are not length-adjusted.
条件
GPT-6 Sol; reasoning effort high. The selected prompts are not representative of typical use.
Responses with at least one factual error4.5%原文 4.5% · 单位 %
公布 未确认
GPT-6 Sol xhigh
设置
GPT-6 Sol; reasoning effort xhigh as directly labeled in the official chart.
未知项
未另记
工具
工具条件未记录
推理预算
xhigh
框架
未记录
协议
High-difficulty anonymized conversations previously flagged by users for factual errors; an answer counts as an error when it contains at least one factual error; scores are not length-adjusted.
条件
GPT-6 Sol; reasoning effort xhigh. The selected prompts are not representative of typical use.
Responses with at least one factual error4.6%原文 4.6% · 单位 %
公布 未确认
GPT-6 Sol max
设置
GPT-6 Sol; reasoning effort max as directly labeled in the official chart.
未知项
未另记
工具
工具条件未记录
推理预算
max
框架
未记录
协议
High-difficulty anonymized conversations previously flagged by users for factual errors; an answer counts as an error when it contains at least one factual error; scores are not length-adjusted.
条件
GPT-6 Sol; reasoning effort max. The selected prompts are not representative of typical use.
GPT-6 Sol · User-flagged factuality, high-difficulty prompts · Anonymized ChatGPT conversations previously flagged by users for factual errors; exact set version and sample size unspecified · Scores and configurations
GPT-6 Sol record for User-flagged factuality, high-difficulty prompts · Anonymized ChatGPT conversations previously flagged by users for factual errors; exact set version and sample size unspecified. 5 score records for this release, including 5 exact values (may include historical records). Includes 5 source locations.
GPT-6 Sol · User-flagged factuality, high-difficulty prompts scores and configurations
Metric and score
Configuration and evaluation conditions
Sources
Responses with at least one factual error11.4%Reported 11.4% · Unit %
Published Unconfirmed
GPT-6 Sol low
Settings
GPT-6 Sol; reasoning effort low as directly labeled in the official chart.
Unknowns
none recorded
Tools
Tool access not recorded
Reasoning budget
low
Framework
not recorded
Protocol
High-difficulty anonymized conversations previously flagged by users for factual errors; an answer counts as an error when it contains at least one factual error; scores are not length-adjusted.
Conditions
GPT-6 Sol; reasoning effort low. The selected prompts are not representative of typical use.
Responses with at least one factual error6.9%Reported 6.9% · Unit %
Published Unconfirmed
GPT-6 Sol medium
Settings
GPT-6 Sol; reasoning effort medium as directly labeled in the official chart.
Unknowns
none recorded
Tools
Tool access not recorded
Reasoning budget
medium
Framework
not recorded
Protocol
High-difficulty anonymized conversations previously flagged by users for factual errors; an answer counts as an error when it contains at least one factual error; scores are not length-adjusted.
Conditions
GPT-6 Sol; reasoning effort medium. The selected prompts are not representative of typical use.
Responses with at least one factual error5.1%Reported 5.1% · Unit %
Published Unconfirmed
GPT-6 Sol high
Settings
GPT-6 Sol; reasoning effort high as directly labeled in the official chart.
Unknowns
none recorded
Tools
Tool access not recorded
Reasoning budget
high
Framework
not recorded
Protocol
High-difficulty anonymized conversations previously flagged by users for factual errors; an answer counts as an error when it contains at least one factual error; scores are not length-adjusted.
Conditions
GPT-6 Sol; reasoning effort high. The selected prompts are not representative of typical use.
Responses with at least one factual error4.5%Reported 4.5% · Unit %
Published Unconfirmed
GPT-6 Sol xhigh
Settings
GPT-6 Sol; reasoning effort xhigh as directly labeled in the official chart.
Unknowns
none recorded
Tools
Tool access not recorded
Reasoning budget
xhigh
Framework
not recorded
Protocol
High-difficulty anonymized conversations previously flagged by users for factual errors; an answer counts as an error when it contains at least one factual error; scores are not length-adjusted.
Conditions
GPT-6 Sol; reasoning effort xhigh. The selected prompts are not representative of typical use.
Responses with at least one factual error4.6%Reported 4.6% · Unit %
Published Unconfirmed
GPT-6 Sol max
Settings
GPT-6 Sol; reasoning effort max as directly labeled in the official chart.
Unknowns
none recorded
Tools
Tool access not recorded
Reasoning budget
max
Framework
not recorded
Protocol
High-difficulty anonymized conversations previously flagged by users for factual errors; an answer counts as an error when it contains at least one factual error; scores are not length-adjusted.
Conditions
GPT-6 Sol; reasoning effort max. The selected prompts are not representative of typical use.