NLP Lineage

11 August 2025 · Reasoning

GLM-4.5

GLM-4.5

GLM team

Date
11 August 2025
What the date marks
arXiv submission
Kind
Model
Organisation
Zhipu AI / Z.ai
Access
MIT open weights
Threads
In the guided tour
Chapter 8. The weights get out
Learning phases
6. Local models and deployment

A milestone: 1 later milestone descends directly from it, and its descendants reach 2 of the 10 threads. How levels are decided

Read the primary source (arxiv.org)

Full summary in progress. Until then, this page shows the milestone's record from the research data.

What it changed

Before it

Open agentic models came mainly from DeepSeek, Qwen and Moonshot.

The problem

Reasoning, coding and agentic skill were usually strong in one model and weak in the others.

What it introduced

An open MoE explicitly targeting all three together, released under MIT.

Why it matters

It widened the open frontier to a fourth serious lab and pushed the unified-capability framing.

What modern models inherit

Open models competing on agentic benchmarks, not just knowledge.

What came next

GLM-5.2 in 2026.

Before and after

The milestones it descends from, and the milestones that descend from it. Work developed alongside it is not counted. Weak links are left out unless you include them.

Before: where it came from

No documented ancestors are recorded for GLM-4.5 without weak links.

No documented ancestors are recorded for GLM-4.5 without weak links.

GLM-4.5 has 72 recorded ancestors, including indirect ones, across 12 chapters, reaching back to Recurrent networks in 1990. Showing its direct parents and the 25 turning points among them.

GLM-4.5 has 72 recorded ancestors, including indirect ones, across 12 chapters, reaching back to Recurrent networks in 1990. Showing all of them.

  1. 2025GLM-4.5This milestone.
  2. 2025DeepSeek-R1GLM-4.5 was influenced by this (indirect influence)Chapter 12. Thinking longer
  3. 2024DeepSeek-V3DeepSeek-R1 built directly on thisChapter 8. The weights get out
  4. 2024o1 Turning pointDeepSeek-R1 was influenced by thisChapter 12. Thinking longer
  5. 2024Test-time computeDeepSeek-R1 was influenced by thisChapter 12. Thinking longer
  6. 2024Llama 3.1 405BDeepSeek-V3 was influenced by thisChapter 8. The weights get out
  7. 2024GPT-4oDeepSeek-V3 was influenced by thisChapter 11. Seeing and hearing
  8. 2024DeepSeek-V2 / MLADeepSeek-V3 built directly on thisChapter 9. Making it run anywhere
  9. 2024Llama 3Llama 3.1 405B built directly on thisChapter 8. The weights get out
  10. 2024GRPODeepSeek-R1 built directly on thisChapter 12. Thinking longer
  11. 2024DeepSeekMoEDeepSeek-V3 built directly on thisChapter 8. The weights get out
  12. 2024DeepSeek LLMGRPO built directly on thisChapter 8. The weights get out
  13. 2023GPT-4VGPT-4o built directly on thisChapter 11. Seeing and hearing
  14. 2023Llama 2 Turning pointDeepSeek LLM was influenced by thisChapter 8. The weights get out
  15. 2023Position interpolationLlama 3.1 405B was influenced by thisChapter 10. Reading more than you remember
  16. 2023Process supervisiono1 was influenced by this (indirect influence)
    GRPO was influenced by this (indirect influence)
    Test-time compute was influenced by this    Chapter 12. Thinking longer
  17. 2023DPOLlama 3.1 405B built directly on thisChapter 6. Teaching models what people want
  18. 2023Grouped-query attention Turning pointDeepSeek-V2 / MLA was influenced by this
    Llama 3.1 405B built directly on this    Chapter 9. Making it run anywhere
  19. 2023Tree of thoughtsTest-time compute was influenced by this (indirect influence)Chapter 12. Thinking longer
  20. 2023GPT-4 Turning pointGPT-4o built directly on thisChapter 7. The assistant goes public
  21. 2023LLaMA Turning pointLlama 2 built directly on this
    Position interpolation built directly on this    Chapter 8. The weights get out
  22. 2022GPT-3.5 Turning pointGPT-4 succeeded thisChapter 7. The assistant goes public
  23. 2022WhisperGPT-4o was influenced by this (indirect influence)Chapter 11. Seeing and hearing
  24. 2022Chinchilla Turning pointDeepSeek LLM was influenced by this
    Llama 3.1 405B was influenced by this    Chapter 5. Scale becomes a science
  25. 2022STaRo1 was influenced by this (indirect influence)
    Test-time compute was influenced by this (indirect influence)    Chapter 12. Thinking longer
  26. 2022Self-consistencyTest-time compute built directly on thisChapter 12. Thinking longer
  27. 2022InstructGPT Turning pointGRPO was influenced by thisChapter 6. Teaching models what people want
  28. 2022Chain of thought Turning pointDeepSeek-R1 was influenced by thisChapter 12. Thinking longer
  29. 2021GopherChinchilla built directly on thisChapter 5. Scale becomes a science
  30. 2021FLANInstructGPT was influenced by thisChapter 6. Teaching models what people want
  31. 2021RoPE Turning pointDeepSeek-V2 / MLA built directly on thisChapter 10. Reading more than you remember
  32. 2021CLIPGPT-4V was influenced by this (indirect influence)Chapter 11. Seeing and hearing
  33. 2021Switch TransformerDeepSeekMoE was influenced by thisChapter 5. Scale becomes a science
  34. 2020Vision TransformerCLIP built directly on thisChapter 11. Seeing and hearing
  35. 2020RLHF for summarisation Turning pointInstructGPT built directly on thisChapter 6. Teaching models what people want
  36. 2020GShard Turning pointDeepSeekMoE built directly on thisChapter 5. Scale becomes a science
  37. 2020GPT-3 Turning pointChain of thought was influenced by thisChapter 5. Scale becomes a science
  38. 2020Scaling laws Turning pointGPT-3 was influenced by this
    DeepSeek LLM built directly on this    Chapter 5. Scale becomes a science
  39. 2019Multi-query attentionDeepSeek-V2 / MLA was influenced by thisChapter 9. Making it run anywhere
  40. 2019T5 Turning pointSwitch Transformer built directly on this
    FLAN was influenced by this    Chapter 4. Pretrain once, use everywhere
  41. 2019C4 Turning pointT5 was influenced by thisChapter 4. Pretrain once, use everywhere
  42. 2019RLHF for LMsRLHF for summarisation built directly on thisChapter 6. Teaching models what people want
  43. 2019Sparse Transformer Turning pointGPT-3 built directly on thisChapter 4. Pretrain once, use everywhere
  44. 2019GPT-2 Turning pointGPT-3 built directly on thisChapter 4. Pretrain once, use everywhere
  45. 2019WebTextGPT-2 was influenced by thisChapter 4. Pretrain once, use everywhere
  46. 2019Transformer-XLSparse Transformer was influenced by this (indirect influence)
    RoPE was influenced by this    Chapter 4. Pretrain once, use everywhere
  47. 2018BERT Turning pointT5 was influenced by thisChapter 4. Pretrain once, use everywhere
  48. 2018SentencePieceT5 built directly on thisChapter 4. Pretrain once, use everywhere
  49. 2018GPT-1GPT-2 built directly on thisChapter 4. Pretrain once, use everywhere
  50. 2018ELMoGPT-1 was influenced by this (indirect influence)Chapter 4. Pretrain once, use everywhere
  51. 2018ULMFiTGPT-1 was influenced by this (indirect influence)Chapter 4. Pretrain once, use everywhere
  52. 2017Mixed precisionDeepSeek-V3 was influenced by thisChapter 9. Making it run anywhere
  53. 2017CoVeELMo was influenced by thisChapter 3. Attention, then only attention
  54. 2017RLHF origino1 was influenced by this (indirect influence)Chapter 6. Teaching models what people want
  55. 2017Transformer Turning pointGPT-2 built directly on this
    Sparse Transformer built directly on this
    Scaling laws built directly on this    Chapter 3. Attention, then only attention
  56. 2017ConvS2STransformer was influenced by this (indirect influence)Chapter 3. Attention, then only attention
  57. 2017Sparse MoE Turning pointDeepSeekMoE was influenced by thisChapter 3. Attention, then only attention
  58. 2016Layer normGPT-2 built directly on thisChapter 3. Attention, then only attention
  59. 2015Residual connectionsTransformer was influenced by thisChapter 3. Attention, then only attention
  60. 2015LM pretraining Turning pointGPT-1 was influenced by thisChapter 2. Words become vectors
  61. 2015BPE tokenizationGPT-2 built directly on thisChapter 2. Words become vectors
  62. 2015Luong attentionTransformer was influenced by thisChapter 3. Attention, then only attention
  63. 2014AdamTransformer was influenced by thisChapter 2. Words become vectors
  64. 2014GloVeELMo built directly on thisChapter 2. Words become vectors
  65. 2014Seq2seq Turning pointTransformer built directly on thisChapter 2. Words become vectors
  66. 2014Attention Turning pointTransformer built directly on thisChapter 3. Attention, then only attention
  67. 2014GRU / encoder-decoderAttention built directly on thisChapter 2. Words become vectors
  68. 2013word2vecELMo was influenced by thisChapter 2. Words become vectors
  69. 2011NLP from scratchELMo was influenced by thisChapter 1. Language as probability
  70. 2003Neural LMNLP from scratch was influenced by this
    word2vec was influenced by this    Chapter 1. Language as probability
  71. 1997LSTM Turning pointSparse MoE built directly on thisChapter 1. Language as probability
  72. 1991Mixture of expertsSparse MoE built directly on thisChapter 1. Language as probability
  73. 1990Recurrent networksLSTM was influenced by thisChapter 1. Language as probability

After: what it led to

GLM-4.5 has 1 documented descendants across 1 chapter, reaching forward to GLM-5.2 in 2026. Showing all of them.

GLM-4.5 has 1 documented descendants across 1 chapter, reaching forward to GLM-5.2 in 2026. Showing all of them.

GLM-4.5 has 1 recorded descendants, including indirect ones, across 1 chapter, reaching forward to GLM-5.2 in 2026. Showing all of them.

GLM-4.5 has 1 recorded descendants, including indirect ones, across 1 chapter, reaching forward to GLM-5.2 in 2026. Showing all of them.

  1. 2025GLM-4.5This milestone.
  2. 2026GLM-5.2built directly on GLM-4.5Chapter 8. The weights get out
  1. 2025GLM-4.5This milestone.
  2. 2026GLM-5.2built directly on GLM-4.5Chapter 8. The weights get out

Direct links and their sources

Where it came from

Influenced by

What it led to

Built on directly by