GDM

17 atoms · 跨 9 天 · 首见 2026-04-08 · 最近 2026-06-23

三色: 🟦 fact 3 · 🟥 take 14 · stance ▲6/▼6/◆3

来源: X 17

时态: fresh:10 · aging:6 · stale:1

标签: 好观点:13 · 好信源:2 · 好数字:1 · 好思考:1

叙事 (narrative) (9)

• 2026-06-16 · long interest in multimodality, robotics and agents in generative worlds
  first from GDM since they've long been interested in multimodality, robotics and agents in generative worlds, and recent works like Genie point in this direction
  tags: 好信源 · → daily
  value: direction=
• 2026-06-16 · compute spending potential
  they *could*, and certainly it'd be 'easy' to spend a similar amount
  tags: 好思考 · → daily
• 2026-06-23 · lacks resolve to close gap
  they (presumably) read all the same Chinese papers and haven't done this yet, so idk
  tags: 好观点 · → daily
• 2026-06-19 · 编程能力失败
  GDM did fail on coding.
  tags: 好观点 · → daily
• 2026-06-16 · compute spend relative to OpenAI
  I don't think GDM spends significantly more compute on developing frontier LLMs than OpenAI does.
  tags: 好观点 · → daily
• 2026-05-08 [老化中] · AI for science leader status
  GDM is still the undisputed leader in AI for science
  tags: 好观点 · → daily
• 2026-05-08 [老化中] · ceding frontier of agency and programming ability and automated AI research to Anthropic and OpenAI
  interesting in light of them ceding the frontier of agency and programming ability (and perhaps automated *AI research*) to Anthropic and OpenAI
  tags: 好观点 · → daily
• 2026-05-08 [老化中] · 战略判断失误
  GDM은 스텝을 잘못 밟았다고 생각
  tags: 好观点 · → daily

预测 (forecast) (2)

• 2026-06-23 · could close gap by revamping post-training
  I'd like to say GDM can close the gap by just revamping their post-training
  tags: 好观点 · → daily
• 2026-06-01 · IPO 时间预期
  basically all frontier labs will likely be public by the end of this year (OpenAI, Anthropic, GDM, Meta)
  tags: 好观点 · → daily
  value: date=2026年底前

立场 (position) (3)

• 2026-06-14 [老化中] · distillation capability
  GDM is best at distillation
  tags: 好观点 · → daily
• 2026-06-14 [老化中] · pre-training capability
  GDM is the absolute best at pre-training
  tags: 好观点 · → daily
• 2026-06-23 [老化中] · gap behind Anthropic: 6 months and growing
  GDM at 6 months and growing
  tags: 好观点 · → daily
  value: qty=6 months · direction=behind and widening

事实 (fact) (3)

• 2026-05-24 · 系统解决了8个更多问题
  this lands harder after GDM's system solved 8 more
  tags: 好数字 · → daily
  value: qty=8 · date=2026-05-25
• 2026-05-08 · 与EVE Online合作
  GDM Partners with EVE Online
  tags: 好信源 · → daily
• 2026-06-13 · pretraining_quality
  GDM *is* good at pretraining as we understand it
  tags: 好观点 · → daily

时间轴 (近 20)

• 2026-06-23 · position · gap behind Anthropic: 6 months and growing · → daily
• 2026-06-23 · forecast · could close gap by revamping post-training · → daily
• 2026-06-23 · narrative · lacks resolve to close gap · → daily
• 2026-06-19 · narrative · 编程能力失败 · → daily
• 2026-06-16 · narrative · long interest in multimodality, robotics and agents in gener · → daily
• 2026-06-16 · narrative · compute spend relative to OpenAI · → daily
• 2026-06-16 · narrative · compute spending potential · → daily
• 2026-06-14 · position · distillation capability · → daily
• 2026-06-14 · position · pre-training capability · → daily
• 2026-06-13 · fact · pretraining_quality · → daily
• 2026-06-01 · forecast · IPO 时间预期 · → daily
• 2026-05-24 · fact · 系统解决了8个更多问题 · → daily
• 2026-05-08 · fact · 与EVE Online合作 · → daily
• 2026-05-08 · narrative · AI for science leader status · → daily
• 2026-05-08 · narrative · ceding frontier of agency and programming ability and automa · → daily
• 2026-05-08 · narrative · 战略判断失误 · → daily
• 2026-04-08 · narrative · 速度向上の順位 · → daily

← 实体目录 · 系统日志