NLP Lineage

25 March 2025 · Reasoning

Gemini 2.5

Gemini 2.5

Google DeepMind

Date
25 March 2025
What the date marks
Model release
Kind
Model
Organisation
Google DeepMind
Access
API only
Other dates
general availability 2025-06-05
Threads
In the guided tour
Chapter 12. Thinking longer
Learning phases
2. How LLMs are trained; 3. API usage and prompting

A milestone: 2 later milestones descend directly from it, and its descendants reach 4 of the 10 threads. How levels are decided

Figures quoted here are reported by the organisation that built it, not independently verified.

Read the primary source (blog.google)

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

What it changed

Before it

Thinking was an experimental mode on one model variant.

The problem

If reasoning helps broadly, it should not be a special case.

What it introduced

Thinking built into every model in the family by default, with long context retained.

Why it matters

It made reasoning table stakes rather than a premium tier, across a whole product line.

What modern models inherit

Reasoning on by default.

What came next

Gemini 3 and the Deep Think variants.

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

Gemini 2.5 has 36 documented ancestors across 9 chapters, reaching back to Recurrent networks in 1990. Showing its direct parents and the 14 turning points among them.

Gemini 2.5 has 36 documented ancestors across 9 chapters, reaching back to Recurrent networks in 1990. Showing all of them.

Gemini 2.5 has 69 recorded ancestors, including indirect ones, across 12 chapters, reaching back to Recurrent networks in 1990. Showing its direct parents and the 20 turning points among them.

Gemini 2.5 has 69 recorded ancestors, including indirect ones, across 12 chapters, reaching back to Recurrent networks in 1990. Showing all of them.

  1. 2025Gemini 2.5This milestone.
  2. 2024Gemini 2.0Gemini 2.5 built directly on thisChapter 14. Where the threads meet now
  3. 2024o1 Turning pointGemini 2.5 was influenced by thisChapter 12. Thinking longer
  4. 2024Gemini 1.5Gemini 2.5 was influenced by thisChapter 10. Reading more than you remember
  5. 2023Gemini 1.0Gemini 1.5 built directly on thisChapter 11. Seeing and hearing
  6. 2022PaLMGemini 1.0 succeeded thisChapter 5. Scale becomes a science
  7. 2022Chain of thought Turning pointo1 was influenced by thisChapter 12. Thinking longer
  8. 2020Vision TransformerGemini 1.0 was influenced by thisChapter 11. Seeing and hearing
  9. 2020GShard Turning pointGemini 1.5 was influenced by thisChapter 5. Scale becomes a science
  10. 2020GPT-3 Turning pointChain of thought was influenced by thisChapter 5. Scale becomes a science
  11. 2020Scaling laws Turning pointGPT-3 was influenced by thisChapter 5. Scale becomes a science
  12. 2019Sparse Transformer Turning pointGPT-3 built directly on thisChapter 4. Pretrain once, use everywhere
  13. 2019GPT-2 Turning pointGPT-3 built directly on thisChapter 4. Pretrain once, use everywhere
  14. 2019WebTextGPT-2 was influenced by thisChapter 4. Pretrain once, use everywhere
  15. 2018BERT Turning pointVision Transformer was influenced by thisChapter 4. Pretrain once, use everywhere
  16. 2018GPT-1BERT was influenced by this
    GPT-2 built directly on this    Chapter 4. Pretrain once, use everywhere
  17. 2018ELMoBERT was influenced by thisChapter 4. Pretrain once, use everywhere
  18. 2018ULMFiTBERT was influenced by thisChapter 4. Pretrain once, use everywhere
  19. 2017CoVeELMo was influenced by thisChapter 3. Attention, then only attention
  20. 2017Transformer Turning pointVision Transformer built directly on thisChapter 3. Attention, then only attention
  21. 2017Sparse MoE Turning pointGShard built directly on thisChapter 3. Attention, then only attention
  22. 2016Layer normTransformer built directly on this
    GPT-2 built directly on this    Chapter 3. Attention, then only attention
  23. 2015Residual connectionsTransformer was influenced by thisChapter 3. Attention, then only attention
  24. 2015LM pretraining Turning pointBERT was influenced by thisChapter 2. Words become vectors
  25. 2015BPE tokenizationGPT-2 built directly on thisChapter 2. Words become vectors
  26. 2015Luong attentionTransformer was influenced by thisChapter 3. Attention, then only attention
  27. 2014AdamTransformer was influenced by thisChapter 2. Words become vectors
  28. 2014GloVeELMo built directly on thisChapter 2. Words become vectors
  29. 2014Seq2seq Turning pointTransformer built directly on thisChapter 2. Words become vectors
  30. 2014Attention Turning pointTransformer built directly on thisChapter 3. Attention, then only attention
  31. 2014GRU / encoder-decoderAttention built directly on thisChapter 2. Words become vectors
  32. 2013word2vecELMo was influenced by thisChapter 2. Words become vectors
  33. 2011NLP from scratchELMo was influenced by thisChapter 1. Language as probability
  34. 2003Neural LMNLP from scratch was influenced by this
    word2vec was influenced by this    Chapter 1. Language as probability
  35. 1997LSTM Turning pointSparse MoE built directly on thisChapter 1. Language as probability
  36. 1991Mixture of expertsSparse MoE built directly on thisChapter 1. Language as probability
  37. 1990Recurrent networksLSTM was influenced by thisChapter 1. Language as probability
  1. 2025Gemini 2.5This milestone.
  2. 2024Gemini 2.0Gemini 2.5 built directly on thisChapter 14. Where the threads meet now
  3. 2024Computer useGemini 2.0 was influenced by this (indirect influence)Chapter 13. Models that act
  4. 2024o1 Turning pointGemini 2.5 was influenced by thisChapter 12. Thinking longer
  5. 2024Claude 3.5 SonnetComputer use built directly on thisChapter 14. Where the threads meet now
  6. 2024Claude 3Claude 3.5 Sonnet built directly on thisChapter 14. Where the threads meet now
  7. 2024Gemini 1.5Gemini 2.5 was influenced by thisChapter 10. Reading more than you remember
  8. 2023Gemini 1.0Gemini 1.5 built directly on thisChapter 11. Seeing and hearing
  9. 2023GPT-4VComputer use was influenced by thisChapter 11. Seeing and hearing
  10. 2023Lost in the middleGemini 1.5 was influenced by this (indirect influence)Chapter 10. Reading more than you remember
  11. 2023Function callingComputer use was influenced by this (indirect influence)Chapter 13. Models that act
  12. 2023Process supervisiono1 was influenced by this (indirect influence)Chapter 12. Thinking longer
  13. 2023100K contextLost in the middle was influenced by this (indirect influence)Chapter 10. Reading more than you remember
  14. 2023GPT-4 Turning pointGPT-4V built directly on thisChapter 7. The assistant goes public
  15. 2023Claude100K context built directly on thisChapter 7. The assistant goes public
  16. 2023ToolformerFunction calling was influenced by this (indirect influence)Chapter 13. Models that act
  17. 2022Constitutional AIClaude built directly on this
    Claude 3 built directly on this    Chapter 6. Teaching models what people want
  18. 2022GPT-3.5 Turning pointGPT-4 succeeded thisChapter 7. The assistant goes public
  19. 2022ReActFunction calling was influenced by this (indirect influence)Chapter 13. Models that act
  20. 2022HH-RLHFClaude was influenced by thisChapter 6. Teaching models what people want
  21. 2022PaLMGemini 1.0 succeeded thisChapter 5. Scale becomes a science
  22. 2022STaRo1 was influenced by this (indirect influence)Chapter 12. Thinking longer
  23. 2022InstructGPT Turning pointGPT-4 built directly on thisChapter 6. Teaching models what people want
  24. 2022Chain of thought Turning pointo1 was influenced by thisChapter 12. Thinking longer
  25. 2021WebGPTReAct was influenced by this
    Toolformer was influenced by this    Chapter 13. Models that act
  26. 2021HHH assistantHH-RLHF built directly on thisChapter 6. Teaching models what people want
  27. 2021FLANInstructGPT was influenced by thisChapter 6. Teaching models what people want
  28. 2021CLIPGPT-4V was influenced by this (indirect influence)Chapter 11. Seeing and hearing
  29. 2020Vision TransformerGemini 1.0 was influenced by thisChapter 11. Seeing and hearing
  30. 2020RLHF for summarisation Turning pointInstructGPT built directly on this
    WebGPT built directly on this
    HH-RLHF built directly on this    Chapter 6. Teaching models what people want
  31. 2020GShard Turning pointGemini 1.5 was influenced by thisChapter 5. Scale becomes a science
  32. 2020GPT-3 Turning pointChain of thought was influenced by thisChapter 5. Scale becomes a science
  33. 2020RAGLost in the middle was influenced by thisChapter 10. Reading more than you remember
  34. 2020Dense retrievalRAG built directly on thisChapter 10. Reading more than you remember
  35. 2020Scaling laws Turning pointGPT-3 was influenced by thisChapter 5. Scale becomes a science
  36. 2019BARTRAG built directly on thisChapter 4. Pretrain once, use everywhere
  37. 2019T5 Turning pointFLAN was influenced by thisChapter 4. Pretrain once, use everywhere
  38. 2019C4 Turning pointT5 was influenced by thisChapter 4. Pretrain once, use everywhere
  39. 2019RLHF for LMsRLHF for summarisation built directly on thisChapter 6. Teaching models what people want
  40. 2019Sparse Transformer Turning pointGPT-3 built directly on thisChapter 4. Pretrain once, use everywhere
  41. 2019GPT-2 Turning pointGPT-3 built directly on thisChapter 4. Pretrain once, use everywhere
  42. 2019WebTextGPT-2 was influenced by thisChapter 4. Pretrain once, use everywhere
  43. 2019Transformer-XLSparse Transformer was influenced by this (indirect influence)Chapter 4. Pretrain once, use everywhere
  44. 2018BERT Turning pointVision Transformer was influenced by thisChapter 4. Pretrain once, use everywhere
  45. 2018SentencePieceT5 built directly on thisChapter 4. Pretrain once, use everywhere
  46. 2018GPT-1BERT was influenced by this
    GPT-2 built directly on this    Chapter 4. Pretrain once, use everywhere
  47. 2018ELMoBERT was influenced by thisChapter 4. Pretrain once, use everywhere
  48. 2018ULMFiTBERT was influenced by thisChapter 4. Pretrain once, use everywhere
  49. 2017CoVeELMo was influenced by thisChapter 3. Attention, then only attention
  50. 2017RLHF origino1 was influenced by this (indirect influence)Chapter 6. Teaching models what people want
  51. 2017Transformer Turning pointVision Transformer built directly on thisChapter 3. Attention, then only attention
  52. 2017ConvS2STransformer was influenced by this (indirect influence)Chapter 3. Attention, then only attention
  53. 2017FAISSRAG was influenced by thisChapter 10. Reading more than you remember
  54. 2017Sparse MoE Turning pointGShard built directly on thisChapter 3. Attention, then only attention
  55. 2016Layer normTransformer built directly on this
    GPT-2 built directly on this    Chapter 3. Attention, then only attention
  56. 2015Residual connectionsTransformer was influenced by thisChapter 3. Attention, then only attention
  57. 2015LM pretraining Turning pointBERT was influenced by thisChapter 2. Words become vectors
  58. 2015BPE tokenizationGPT-2 built directly on thisChapter 2. Words become vectors
  59. 2015Luong attentionTransformer was influenced by thisChapter 3. Attention, then only attention
  60. 2014AdamTransformer was influenced by thisChapter 2. Words become vectors
  61. 2014GloVeELMo built directly on thisChapter 2. Words become vectors
  62. 2014Seq2seq Turning pointTransformer built directly on thisChapter 2. Words become vectors
  63. 2014Attention Turning pointTransformer built directly on thisChapter 3. Attention, then only attention
  64. 2014GRU / encoder-decoderAttention built directly on thisChapter 2. Words become vectors
  65. 2013word2vecELMo was influenced by thisChapter 2. Words become vectors
  66. 2011NLP from scratchELMo was influenced by thisChapter 1. Language as probability
  67. 2003Neural LMNLP from scratch was influenced by this
    word2vec was influenced by this    Chapter 1. Language as probability
  68. 1997LSTM Turning pointSparse MoE built directly on thisChapter 1. Language as probability
  69. 1991Mixture of expertsSparse MoE built directly on thisChapter 1. Language as probability
  70. 1990Recurrent networksLSTM was influenced by thisChapter 1. Language as probability

After: what it led to

Gemini 2.5 has 5 documented descendants across 2 chapters, reaching forward to Gemini 3.1 in 2026. Showing its direct descendants.

Gemini 2.5 has 5 documented descendants across 2 chapters, reaching forward to Gemini 3.1 in 2026. Showing all of them.

Gemini 2.5 has 5 recorded descendants, including indirect ones, across 2 chapters, reaching forward to Gemini 3.1 in 2026. Showing its direct descendants.

Gemini 2.5 has 5 recorded descendants, including indirect ones, across 2 chapters, reaching forward to Gemini 3.1 in 2026. Showing all of them.

  1. 2025Gemini 2.5This milestone.
  2. 2025IMO goldbuilt directly on Gemini 2.5Chapter 12. Thinking longer
  3. 2025IMO-ProofBenchwas influenced by IMO goldChapter 12. Thinking longer
  4. 2025Gemini 3built directly on Gemini 2.5Chapter 14. Where the threads meet now
  5. 2026Gemini 3 Deep Thinkbuilt directly on Gemini 3
    was influenced by IMO gold    Chapter 12. Thinking longer
  6. 2026Gemini 3.1built directly on Gemini 3Chapter 14. Where the threads meet now
  1. 2025Gemini 2.5This milestone.
  2. 2025IMO goldbuilt directly on Gemini 2.5Chapter 12. Thinking longer
  3. 2025IMO-ProofBenchwas influenced by IMO goldChapter 12. Thinking longer
  4. 2025Gemini 3built directly on Gemini 2.5Chapter 14. Where the threads meet now
  5. 2026Gemini 3 Deep Thinkbuilt directly on Gemini 3
    was influenced by IMO gold    Chapter 12. Thinking longer
  6. 2026Gemini 3.1built directly on Gemini 3Chapter 14. Where the threads meet now

Direct links and their sources

Where it came from

Built directly on

Influenced by

What it led to

Built on directly by