专题
返回时间轴
2024-07-23

Meta releases Llama 3.1 405B, the first frontier-level open-source AI model

Capability Breakthrough

事件摘要

Llama 3.1 405B (July 2024) was the first open-weight AI model to reach frontier capabilities. Meta proved open models could match proprietary systems at scale.

影响评估

  • Paradigm Shift +3 · Long-term

    First open-weight model to match closed-source frontier performance. Proved that open-source AI could compete at the highest level. Reset industry debate from 'if' open models could match proprietary to 'when and how.' Accelerated the open-weight ecosystem that defined the 2025 AI landscape.

    Affected Groups: AI developers, open-source community, startups, AI labs

  • Access Democratization +2 · Long-term

    Made frontier-level AI performance available to anyone who could run or host the model. Enabled fine-tuning, customization, and local deployment that proprietary models did not allow. Available in 8B, 70B, and 405B sizes for different compute tiers.

    Affected Groups: developers, researchers, startups, enterprises, global south

共识度与来源

重要度 L2
分类 Capability Breakthrough
共识度 Broad Consensus
影响指数 7/10

关联事件

← Origins
2023-02-24 L2

Meta releases LLaMA, triggering the open-weight LLM revolution

Meta published LLaMA (Large Language Model Meta AI) in February 2023, a set of foundation language models from 7B to 65B parameters trained on publicly available data only. LLaMA-13B outperformed GPT-3 (175B) on most benchmarks, proving that smaller, well-trained models could rival much larger ones. Although originally released under a non-commercial research license, the weights were leaked and quickly fine-tuned by the community — spawning the open-weight LLM ecosystem that reshaped the AI industry.

2023-03-14 L2

OpenAI releases GPT-4, the first multimodal large language model

On March 14, 2023, OpenAI released GPT-4 — a multimodal model that could accept both text and image inputs while generating text outputs. GPT-4 scored in the 90th percentile on the bar exam, the 99th percentile on the SAT, and demonstrated dramatically improved reasoning, factuality, and steerability over GPT-3.5. It was integrated into ChatGPT Plus as a premium tier, and its API became the default for developers building on LLMs. GPT-4 raised the bar on what LLMs could do and set a new baseline for AI capability that persisted until the model release wave of late 2024.

→ Successors
2025-01-20 L2

DeepSeek releases R1, an open-source reasoning model that shocks global markets

On January 20, 2025, Chinese AI lab DeepSeek released R1, an open-source reasoning model matching OpenAI's o1 on key benchmarks at a fraction of the training cost. Using Mixture-of-Experts architecture and Group Relative Policy Optimization, R1 achieved o1-level reasoning for approximately 1/20th of the training compute. On January 27, the release triggered a global stock market selloff that erased $593 billion from Nvidia's market cap — the largest single-day value loss in history — and DeepSeek became the most-downloaded free app on the US iOS App Store.

2025-04-01 L1

Meta releases Llama 4, the first natively multimodal open-weight model family

In April 2025, Meta released Llama 4, the first open-weight model family built with natively multimodal architecture — processing text, images, and video jointly from the ground up rather than adding vision as a separate module. The family included Llama 4 Maverick (flagship reasoning) and Llama 4 Scout (efficient deployment). Building on the Llama 3.1 405B breakthrough, Llama 4 established open-weight multimodal AI as the new baseline for the open-source ecosystem.

Local Lineage Meta releases Llama ... 2024-07-23 · L2 Meta releases LLaM... 2023-02-24 · L2 OpenAI releases GP... 2023-03-14 · L2 DeepSeek releases ... 2025-01-20 · L2 Meta releases Llam... 2025-04-01 · L1