AI Tech: 2024 Review
...explained with a visual summary.
As we near the end of 2024, it’s time to reflect on how the AI landscape has evolved over the last 12 months.
Below, let’s look at the month-by-month highlights of 2024 in AI technology.
January
OpenAI launches ChatGPT Store for custom AI agents: OpenAI introduced the ChatGPT Store, allowing users to create and share custom AI agents built on GPT-4.
Meta releases Code Llama 70B: Meta launched one of the largest open-source AI models for code generation, Code Llama 70B, enabling developers to generate accurate and complex code across multiple programming languages.
February
OpenAI unveils Sora for video generation: OpenAI introduced Sora, a groundbreaking tool for generating hyper-realistic videos up to a minute from simple text prompts.
Google introduces Gemini 1.5: Google has rebranded Bard to Gemini and released Gemini 1.5, its advanced AI model showcasing enhanced performance and a 1M token context window, the largest ever.
Mistral releases Mistral Large: Mistral AI released its highly anticipated model, Mistral Large, a cutting-edge text generation model designed for complex multilingual reasoning tasks.
Stability AI announces Stable Diffusion 3: Stability AI released Stable Diffusion 3 (SD3), a text-to-image generation model built on a novel "diffusion transformer" architecture. SD3 offers significant image quality, accuracy, and efficiency improvements over its predecessors.
March
xAI open-sources Grok to challenge ChatGPT: Elon Musk’s AI startup, xAI, launched Grok, a ChatGPT competitor, as open-source software. Designed to offer real-time information access and “unrestricted viewpoints,”
Claude 3 redefines GenAI: Anthropic’s Claude 3 series raised the bar for AI performance, excelling in reasoning, mathematics, and coding and outperforming OpenAI’s GPT-4 in several benchmarks.
Cognition Labs launches Devin, an autonomous software engineer: Cognition Labs introduced Devin, an advanced autonomous AI designed to revolutionize software development.
April
Meta’s breakthroughs in GenAI with open-source Llama 3: Meta introduced LLaMA 3, a major upgrade to its large language model family that promises faster and more accurate responses.
OpenAI expands ChatGPT with memory: OpenAI rolled out a groundbreaking memory feature for ChatGPT, enabling the AI to recall user preferences and prior conversations.
ElevenLabs introduces professional voice cloning: ElevenLabs released its Professional Voice Cloning service, enabling users to create near-perfect digital replicas of their voices.
May
OpenAI announces GPT-4o ("Omni") for versatile AI applications: OpenAI launched GPT-4o, nicknamed "Omni," a cutting-edge iteration of its GPT family.
Microsoft releases Phi-3 Vision with advanced predictive capabilities. Microsoft introduced Phi-3, an AI model emphasizing real-time predictive analytics and contextual reasoning.
June
Apple announces Apple Intelligence: Apple announced significant advancements in its AI ecosystem, branded as “Apple Intelligence." The ecosystem is on-device and integrated with apps like Siri, Photos, and Notes.
Anthropic launches Claude 3.5 Sonnet: Anthropic unveiled Claude 3.5 Sonnet, a significant upgrade in its Claude AI series. It improves contextual understanding and the ability to handle lengthy conversations without losing coherence.
Runway launches Gen-3 Alpha for video generation: Runway released Gen-3 Alpha, a cutting-edge generative AI model for video creation, which offers enhanced realism and seamless video transitions, allowing creators to produce high-quality content with minimal effort.
Yandex releases an open-source LLM optimization toolkit: Yandex introduced YaFSDP, an open-source framework designed to significantly reduce GPU usage when deploying large language models, competing with prevailing techniques like FSDP and DeepSpeed.
July
OpenAI introduces SearchGPT for AI-powered search: OpenAI launched SearchGPT, designed to revolutionize how users interact with search engines.
Mistral AI releases Mistral Large 2: Mistral AI released its second-generation language model, Mistral Large 2, featuring significant improvements in multilingual reasoning and computational efficiency.
Meta launches Llama 3.1: Meta introduced Llama 3.1, the largest-ever open-source AI model to date.
OpenAI reveals GPT-4o Mini for efficiency and affordability: OpenAI made GPT-4o Mini available, targeting users and businesses needing cost-effective AI solutions without compromising performance.
August
MiMicrosoft'shi-3.5 models set a new benchmark: Microsoft introduced its Phi-3.5 models, which deliver state-of-the-art performance across natural language processing and understanding tasks and outperform offerings from Google and OpenAI in benchmarks.
Google added new features to Gemini: Google expanded GeGemini's capabilities with two major updates. The first, Gems, allows users to create and personalize AI experts tailored to specific tasks or interests, enabling highly customizable AI interactions. The second is Imagen 3, a cutting-edge image generation model that produces realistic visuals and creative artwork with improved precision and detail.
September
Hugging Face hits 1 Million AI models milestone: Hugging Face achieved a significant milestone, hosting over one million AI models on its platform.
OpenAI releases o1 and o1-mini models: OpenAI introduced the o1 and o1-mini models, focusing on advanced reasoning capabilities and computational efficiency.
Mistral debuted Pixtral 12B, a large language model that emphasizes multilingual support and nuanced reasoning, to challenge ChatGPT.
October
OpenAI launches ChatGPT Search: OpenAI entered the search engine race with ChatGPT Search, integrating generative AI capabilities into search functionality.
Mistral AI presents Ministral 3B and 8B models for edge computing: Mistral AI released two efficient models, Ministral 3B and 8B, tailored for edge computing environments.
OpenAI introduces ‘S’arm’ ‘or multi-agent collaboration: OpenAI unveiled Swarm, an educational framework designed to enable multiple AI agents to collaborate effectively on complex tasks.
ChatGPT Canvas redefines collaborative work for writing and coding: OpenAI released Canvas, a new tool integrated with ChatGPT to streamline collaborative writing and coding.
November
Mistral introduces Pixtral Large and upgrades Le Chat: Mistral AI announced Pixtral Large, a cutting-edge large language model with enhanced multilingual reasoning, coding capabilities, and processing efficiency. In addition, Mistral upgraded its free conversational AI, Le Chat, adding new features and refining user interaction.
OpenAI acquires Chat.com: OpenAI expanded its portfolio by acquiring Chat.com, a strategic move to strengthen its direct consumer offerings.
December
Meta introduces Llama 3.3: Meta expanded its Llama model lineup with Llama 3.3, their latest iteration of the open-weight language model series, featuring enhanced reasoning capabilities and multimodal support.
OpenAI reveals O3 with “AI-like” capabilities: OpenAI announced its latest AI model, O3, which focuses on enhanced reasoning capabilities and outperforms previous models in complex tasks. Interestingly, it only took three months for OpenAI to go from O1 to O3.
Google introduces Veo 2: Google releases an advanced AI video generator called Veo 2, which produces higher-quality, longer-duration videos with improved realism. Early comparisons indicate that Veo 2 performs superiorly to OpenAI Sora due to Google's extensive use of YouTube data for training.
Cohere releases Command R-7B for Cohere's Command R-7B model, which focuses on retrieval-augmented generation, making it ideal for enterprise search and document analysis.
As you can tell, 2024 solidified AI’s role as a transformative technology compared to 2023, and it has set a solid stage for further advancements in 2025.
As we approach 2025, the year's lessons and achievements must set a strong foundation for continued progress and responsible AI development.
👉 Over to you: What are your AI predictions for 2025?
Let us know in the comments.
Thanks for reading.