Timeline of Meta AI/Llama

From Timelines
Jump to navigation Jump to search
Date Event type Details
2013 Founding Facebook founds Facebook Artificial Intelligence Research (FAIR), the predecessor to Meta AI, with Yann LeCun as its head.[1][2]
2016 Partnership FAIR joins Google, Amazon (company), IBM, and Microsoft in founding the Partnership on AI.[3]
2017 Release FAIR publicly releases PyTorch, an open-source machine learning framework building on the earlier Torch (machine learning) modules; it is later adopted by Tesla, Inc.'s Autopilot and Uber's Pyro project.[4][5]
2017 Controversy Media outlets falsely report that FAIR shut down two chatbots after they began "speaking" in a language unintelligible to humans; FAIR clarifies the project was ended because its initial research goal had been met, not out of fear.[6][7]
2018-01-23 Leadership Jérôme Pesenti, formerly Chief technology officer of IBM's big data group, succeeds Yann LeCun as head of Meta AI (then FAIR).[8]
2021-10-28 Rebranding Following Facebook, Inc.'s rebrand to Meta Platforms, FAIR is renamed Meta AI.[9]
2022-11-15 Release/Withdrawal Meta AI releases Galactica (language model), a large language model for generating scientific text; it is withdrawn after three days for producing racist and inaccurate content.[10][11]
2023-02-24 Release Meta AI announces and releases Llama (language model) (Llama 1), a family of foundation large language models ranging from 6.7B to 65.2B parameters, trained on 1–1.4 trillion tokens; weights are initially restricted to approved researchers under a non-commercial license.[12][13][14]
2023-03-03 Leak Llama's model weights leak online via a BitTorrent link posted to 4chan, spreading rapidly through AI communities despite Meta's restricted-access policy; Meta subsequently files takedown requests against HuggingFace and GitHub mirrors.[15][16]
2023-03-10 Release Developer Georgi Gerganov releases llama.cpp, an open-source C++ reimplementation of Llama enabling inference on consumer hardware without a powerful GPU; the project later introduces the GGUF file format.[17]
2023-03-13 Release/Withdrawal Stanford University's Institute for Human-Centered AI releases Alpaca, an instruction-tuned fine-tune of Llama 7B; the model files are taken down on March 21, 2023 over hosting costs and safety concerns, though code and paper remain available.[18][19]
2023-07-18 Release Meta releases Llama (language model)#Llama 2 in partnership with Microsoft, in 7B, 13B, and 70B parameter sizes, trained on 2 trillion tokens with foundation and chat-tuned variants; unlike Llama 1, weights are released for broad commercial use subject to an acceptable use policy.[20][21]
2023-07-20 Criticism Open Source Initiative head Stefano Maffulli states Llama 2's license does not meet the Open Source Definition, accusing Meta of openwashing the model.[22][23]
2023-08-24 Release Meta releases Code Llama, a fine-tune of Llama 2 on code-specific datasets, in 7B, 13B, and 34B sizes; a 70B version follows on January 29, 2024.[24][25]
2023-12-01 Release Meta AI unveils Seamless, a set of natural language processing models for real-time voice-to-voice translation.[26]
2024-04-18 Release Meta releases Llama (language model)#Llama 3 (8B and 70B), pre-trained on roughly 15 trillion tokens; Meta's testing shows the 70B model outperforming Gemini (chatbot) Pro 1.5 and Claude (language model) 3 Sonnet on most benchmarks. Meta simultaneously rolls out the Meta AI virtual assistant, built on Llama, to Facebook and WhatsApp.[27][28]
2024-05 Controversy Meta AI's chatbot begins summarizing news content from outlets without linking to original articles, including in Canada where news links are banned on Meta's platforms, raising attribution and compensation concerns.[29]
2024-07-23 Release Meta releases Llama 3.1 in 8B, 70B, and 405B parameter sizes, trained on 15 trillion tokens with a 128,000-token context window.[30][31]
2024-09-25 Release Meta releases Llama 3.2 in 1B, 3B, 11B, and 90B sizes, its first Llama generation with image-processing (multimodal) capability.[32][33]
2024-10-28 Criticism The Open Source Initiative publishes the Open Source AI Definition (OSAID), requiring disclosure of training data details that Meta does not provide for Llama; a Nature (journal) article separately describes Llama 3 as a case of "openwashing" a closed system.[34][35]
2024-11-01 Controversy Reuters and the Jamestown Foundation report that researchers at China's People's Liberation Army Academy of Military Sciences developed a military-oriented tool based on Llama, which Meta says was unauthorized under Llama's license terms prohibiting military use.[36][37]
2024-11-17 Policy change Meta grants the United States Government and US military contractors permission to use Llama for defense applications, while continuing to prohibit military use by non-US entities.[38][39]
2024-12-07 Release Meta releases Llama 3.3, a 70B-parameter model.[40]
2025-01-09 Legal In a copyright lawsuit brought by authors including Richard Kadrey, Sarah Silverman, and Ta-Nehisi Coates, filings allege CEO Mark Zuckerberg authorized training Llama models on copyrighted content from Library Genesis and concealed the practice by removing copyright markers from the data.[41]
2025-02-18 Criticism The Open Source Initiative publishes a follow-up post reiterating that Meta's Llama license still does not meet its open-source standard.[42]
2025-04-05 Release Meta releases Llama (language model)#Llama 4 Scout (109B total/17B active parameters, 10M-token context) and Maverick (400B total/17B active parameters, 1M-token context), its first natively multimodal, mixture-of-experts Llama generation; the larger unreleased Behemoth model (~2T parameters) is announced as still in training.[43][44]
2025-04-08 Controversy Meta faces criticism after it emerges that Llama 4 Maverick's benchmark score claiming to beat GPT-4o on LMArena was achieved using an unreleased, conversationally-optimized experimental chat version rather than the publicly released model; LMArena says Meta's interpretation of its policy "did not match what we expect from model providers" and revises its policies in response.[45][46][47]
2025-10-01 Policy change Meta announces it will begin using people's interactions with Meta AI to personalize content and advertising shown across its apps.[48][49]
2025-12-22 Controversy French outlet Mediapart reports that in 2022, Meta illegally used works aggregated by the pirate site Library Genesis to train its AI models, describing the role of a Meta employee who later co-founded Mistral AI.[50]
2026-04-09 Discontinuation Meta Superintelligence Labs releases Muse Spark, which replaces Llama as Meta's flagship AI model line.[51]
  1. ↑ "NYU "Deep Learning" Professor LeCun Will Head Facebook's New Artificial Intelligence Lab". TechCrunch. 9 December 2013. Retrieved 2022-05-08.
  2. ↑ "Facebook's AI team hires Vladimir Vapnik, father of the popular support vector machine algorithm". VentureBeat. 2014-11-25.
  3. ↑ Parloff, Roger (September 28, 2016). "AI Partnership Launched by Facebook, Google, Amazon, Microsoft, and IBM". Fortune. Retrieved 2026-06-05.
  4. ↑ "FAIR turns five: What we've accomplished and where we're headed". Engineering at Meta. 2018-12-05. Archived from the original on 2022-05-11. Retrieved 2022-05-08.
  5. ↑ Karpathy, Andrej (6 November 2019). "PyTorch at Tesla - Andrej Karpathy, Tesla". YouTube. Archived from the original on 2023-03-24.
  6. ↑ "Facebook researchers shut down AI bots that started speaking in a language unintelligible to humans". Tech2. 2017-07-31. Archived from the original on 2022-05-08. Retrieved 2022-05-08.
  7. ↑ McKay, Tom (2017-08-01). "No, Facebook Did Not Panic and Shut Down an AI Program That Was Getting Dangerously Smart". Gizmodo. Retrieved 2025-05-27.
  8. ↑ Dave, Greshgorn (January 23, 2018). "The head of Facebook's AI research is stepping into a new role as it shakes up management". Quartz. Archived from the original on 2022-05-08. Retrieved May 7, 2022.
  9. ↑ Murphy Kelly, Samantha (October 29, 2021). "Facebook changes its company name to Meta". CNN Business. Archived from the original on 2022-05-07. Retrieved May 7, 2022.
  10. ↑ "Why Meta's latest large language model survived only three days online". MIT Technology Review. Retrieved 2025-07-18.
  11. ↑ Edwards, Benj (18 November 2022). "New Meta AI demo writes racist and inaccurate scientific literature, gets pulled". Ars Technica. Retrieved 30 December 2022.
  12. ↑ Leswing, Kif (2023-02-24). "Mark Zuckerberg announces Meta's new large language model as A.I. race heats up". CNBC. Retrieved 2025-04-14.
  13. ↑ "Introducing LLaMA: A foundational, 65-billion-parameter large language model". Meta AI. 24 February 2023. Archived from the original on 2023-03-03. Retrieved 16 March 2023.
  14. ↑ Template:Cite arxiv
  15. ↑ Vincent, James (8 March 2023). "Meta's powerful AI language model has leaked online — what happens now?". The Verge. Archived from the original on 3 November 2023. Retrieved 16 March 2023.
  16. ↑ Hern, Alex (2023-03-07). "TechScape: Will Meta's massive leak democratise AI – and at what cost?". The Guardian. Retrieved 2025-04-10.
  17. ↑ Edwards, Benj (2023-03-13). "You can now run a GPT-3-level AI model on your laptop, phone, and Raspberry Pi". Ars Technica. Archived from the original on 2024-01-09. Retrieved 2024-01-04.
  18. ↑ Taori, Rohan (13 March 2023). "Alpaca: A Strong, Replicable Instruction-Following Model". Stanford Center for Research on Foundation Models. Archived from the original on 6 April 2023.
  19. ↑ Quach, Katyanna. "Stanford takes costly, risky Alpaca AI model offline". The Register.
  20. ↑ "Meta and Microsoft Introduce the Next Generation of LLaMA". Meta. 18 July 2023. Archived from the original on 14 September 2023. Retrieved 21 July 2023.
  21. ↑ Edwards, Benj (2023-07-18). "Meta launches LLaMA-2, a source-available AI model that allows commercial applications". Ars Technica. Archived from the original on 2023-11-07. Retrieved 2023-08-08.
  22. ↑ Maffulli, Stefano (20 July 2023). "Meta's LLaMa license is not Open Source". Open Source Initiative. Retrieved 10 July 2025.
  23. ↑ Chan, Rosalie; Hays, Kali (19 July 2023). "Meta's new 'open source' Llama 2 AI model isn't so open after all". Business Insider. Retrieved 10 July 2025.
  24. ↑ "Introducing Code Llama, a state-of-the-art large language model for coding". ai.meta.com. Archived from the original on 2024-09-27.
  25. ↑ Template:Cite arxiv
  26. ↑ Nuñez, Michael (2023-12-01). "Meta AI unveils 'Seamless' translator for real-time communication across languages". VentureBeat. Archived from the original on 2025-07-24. Retrieved 2026-06-05.
  27. ↑ "Introducing Meta Llama 3: The most capable openly available LLM to date". ai.meta.com. April 18, 2024. Archived from the original on 2024-05-15. Retrieved 2024-04-21.
  28. ↑ Heath, Alex (2024-04-18). "Meta's battle with ChatGPT begins now". The Verge. Retrieved 2025-04-10.
  29. ↑ "Meta walked away from news. Now the company's using it for AI content". The Washington Post. 21 May 2024. Archived from the original on 21 May 2024. Retrieved 22 May 2024.
  30. ↑ "Meta's Llama 3.1 is open-source, kind of. Here's how it could reshape the AI race". Fast Company. 2024-07-23.
  31. ↑ Template:Cite arxiv
  32. ↑ Robison, Kylie (2024-09-25). "Meta releases its first open AI model that can process images". The Verge. Retrieved 2024-09-25.
  33. ↑ Wiggers, Kyle (2024-09-25). "Meta's Llama AI models get multimodal". TechCrunch. Archived from the original on 2024-09-25. Retrieved 2024-09-25.
  34. ↑ Davies, Pascale (28 October 2024). "Why Meta's 'open source' AI isn't all it seems". Euronews. Retrieved 10 July 2025.
  35. ↑ Widder, David Gray; Whittaker, Meredith; West, Sarah Myers (27 November 2024). "Why 'open' AI systems are actually closed, and why this matters". Nature. 635 (8040): 827–833. doi:10.1038/s41586-024-08141-1.
  36. ↑ Cheung, Sunny (October 31, 2024). "PRC Adapts Meta's Llama for Military and Security AI Applications". Jamestown Foundation. Retrieved 2024-11-03.
  37. ↑ Pomfret, James; Pang, Jessie (November 1, 2024). "Chinese researchers develop AI model for military use on back of Meta's Llama". Reuters. Retrieved November 1, 2024.
  38. ↑ Thomas, Prasanth Aby (5 November 2024). "Meta offers Llama AI to US government for national security". CIO. Retrieved 9 December 2024.
  39. ↑ Smith, Matthew S. (17 November 2024). "Meta Opens Its AI Model for the U.S. Military". IEEE Spectrum. Retrieved 9 December 2024.
  40. ↑ "Llama Models". www.llama.com. Archived from the original on April 9, 2025. Retrieved April 20, 2025.
  41. ↑ Wiggers, Kyle (January 9, 2025). "Mark Zuckerberg gave Meta's Llama team the OK to train on copyrighted works, filing claims". Techcrunch. Retrieved January 12, 2025.
  42. ↑ Maris, Jordan (18 February 2025). "Meta's LLaMa license is still not Open Source". Open Source Initiative. Retrieved 10 July 2025.
  43. ↑ "The Llama 4 herd: The beginning of a new era of natively multimodal AI innovation". ai.meta.com. Archived from the original on 2025-04-05. Retrieved 2025-04-05.
  44. ↑ "meta-llama/Llama-4-Maverick-17B-128E · Hugging Face". huggingface.co. 2025-04-05. Retrieved 2025-04-06.
  45. ↑ Robison, Kylie (8 April 2025). "Meta got caught gaming AI benchmarks". The Verge. Retrieved 8 April 2025.
  46. ↑ Wiggers, Kyle (6 April 2025). "Meta's benchmarks for its new AI models are a bit misleading". TechCrunch. Retrieved 8 April 2025.
  47. ↑ Franzen, Carl (8 April 2025). "Meta defends Llama 4 release against 'reports of mixed quality,' blames bugs". VentureBeat. Retrieved 8 April 2025.
  48. ↑ "Improving Your Recommendations on Our Apps With AI at Meta". Meta Newsroom. 1 October 2025. Retrieved 4 June 2026.
  49. ↑ Christoffel, Ryan (2025-10-01). "Meta will use your chats with AI to sell hyper-targeted ads". 9to5Mac. Archived from the original on 2025-12-18. Retrieved 2026-06-05.
  50. ↑ "Comment un cofondateur de Mistral AI a piraté des millions de livres quand il travaillait chez Meta". Mediapart (in français). December 22, 2025. Retrieved December 23, 2025.
  51. ↑ "Meta debuts new AI model in first test of costly 'superintelligence' team". The Guardian. 2026-04-09. Retrieved 2026-04-11.
Retrieved from "https://timelines.issarice.com/index.php?title=Timeline_of_Meta_AI/Llama&oldid=91476"