以下部分引用 AI 总结，现阶段 AI 仍然有幻觉。请以内容的原文为准。
数据新鲜度群聊 07-14 ✓卖方 07-14 ✓Wrap 断档15天 ✕News 07-15 ✓X 断档12天 ✕页面生成 07-15
FrontierMath
10 atoms · 跨 5 天 · 首见 2026-04-28 · 最近 2026-05-13
三色: 🟦 fact 9 · 🟥 take 1 · stance ▲3/▼0/◆0
来源: X 10
时态: fresh:10
标签: 好数字:4 · 好信源:4 · 好观点:2 · 好思考:1
叙事 (narrative) (1)
- 2026-05-13 · minor errors fixability
it's good it's only minor errors and that they are fixable
tags: 好观点 · → daily
预测 (forecast) (1)
- 2026-05-11 · will release updated scores on a corrected dataset after completing a thorough human review
We will release updated scores on a corrected dataset after completing a thorough human review.
tags: 好思考 · → daily
事实 (fact) (8)
- 2026-05-13 · flagged errors confirmed by mathematicians
mathematicians have confirmed the vast majority of flagged errors
tags: 好信源 · → daily
- 2026-05-13 · flagged error severity
most of them seem to be minor issues
tags: 好信源 · → daily
- 2026-04-28 · GPT-5.5 solved previously unsolved Tier 4 problems
Across runs, it and GPT-5.5 solved two Tier 4 problems that no model had solved before
tags: 好数字·好观点 · → daily
value: qty=2 · date=2026-04-28
- 2026-05-11 · AI-assisted review flagged fatal errors in about a third of problems
This has flagged fatal errors in about a third of problems, and we believe most of these flags to be valid.
tags: 好数字 · → daily
value: qty=about a third · date=2026-05-12
- 2026-05-08 · 是否公开
FrontierMath isn't public
tags: 好信源 · → daily
- 2026-05-05 · 招募研讨会参与者
Join us for in-person workshops to develop problems for FrontierMath: Open Problems! We are seeking highly interesting unsolved problems from research mathematics whose solutions can be verified programmatically.
tags: 好信源 · → daily
- 2026-04-28 · GPT-5.5 Pro Tiers 1-3 score
GPT-5.5 Pro also set new records on FrontierMath, scoring 52% on Tiers 1-3 (up from 50%)
tags: 好数字 · → daily
value: qty=52% · date=2026-04-28
- 2026-04-28 · GPT-5.5 Pro Tier 4 score
and 40% on Tier 4 (up from 38%)
tags: 好数字 · → daily
value: qty=40% · date=2026-04-28
时间轴 (近 20)
- 2026-05-13 ·
fact · flagged errors confirmed by mathematicians · → daily
- 2026-05-13 ·
fact · flagged error severity · → daily
- 2026-05-13 ·
narrative · minor errors fixability · → daily
- 2026-05-11 ·
fact · AI-assisted review flagged fatal errors in about a third of · → daily
- 2026-05-11 ·
forecast · will release updated scores on a corrected dataset after com · → daily
- 2026-05-08 ·
fact · 是否公开 · → daily
- 2026-05-05 ·
fact · 招募研讨会参与者 · → daily
- 2026-04-28 ·
fact · GPT-5.5 Pro Tiers 1-3 score · → daily
- 2026-04-28 ·
fact · GPT-5.5 Pro Tier 4 score · → daily
- 2026-04-28 ·
fact · GPT-5.5 solved previously unsolved Tier 4 problems · → daily
← 实体目录 · 系统日志