Track the evolution
of artificial intelligence.
Explore AI models, agents, companies, releases, comparisons and the lineages connecting them — one interconnected record, updated as the field moves.
Moving right now
What shipped recently
Claude Sonnet 4.5
Anthropic's coding- and agent-focused flagship in the Claude 4 family.
GPT-5
OpenAI's next-generation flagship model succeeding GPT-4o.
Grok 4
xAI's Grok 4 reasoning flagship.
Llama 4
Meta's Llama 4 MoE family with very long context and multimodal input.
Gemini 2.5 Pro
Reasoning-focused member of the Gemini 2.5 family.
DeepSeek R1
Open-weight reasoning model that popularized reinforcement-learned chain-of-thought at scale.
DeepSeek V3
671B MoE open-weight base model.
OpenAI o1
OpenAI o1 is a reasoning model that thinks through a problem internally before replying, trading latency and cost for much higher accuracy on maths, science and difficult programming tasks. It reached general availability in December 2024 and set the template for later reasoning models.
Model families & generations
Claude
Anthropic's Claude family of assistant models emphasizing safety and long context.
DeepSeek
DeepSeek's open-weight reasoning and coding model line.
Gemini
Google DeepMind's multimodal Gemini family, successor to PaLM and Bard.
GPT
OpenAI's Generative Pre-trained Transformer family, from GPT-1 through GPT-5.
Grok
xAI's Grok family of conversational models.
Llama
Meta's open-weight Llama family of foundation models.
Mistral
Mistral AI's open-weight and Mixtral MoE models.
OpenAI o1
OpenAI's reasoning-first model line, which spends inference-time compute on a private chain of thought before answering.
Head-to-head
OpenAI o1 vs GPT-5
o1 was the first dedicated reasoning model; GPT-5 folds that reasoning into a general-purpose model that is faster, cheaper and stronger almost everywhere.
vsClaude Sonnet 4.5 vs GPT-4o
A 2025 coding-focused model against the 2024 fast multimodal workhorse: Sonnet 4.5 is far stronger on software work, GPT-4o is cheaper and quicker for everyday chat.
vsLlama 4 vs DeepSeek V3
The two most-deployed open-weight families of 2025: Llama 4 offers an enormous context window and a broad ecosystem, DeepSeek V3 delivers strong reasoning per dollar.
Assistants & autonomous systems
News
Anthropic ships Claude Sonnet 4.5
Anthropic releases Claude Sonnet 4.5 with a focus on coding and agentic workflows.
OpenAI releases GPT-5
OpenAI ships GPT-5, its next-generation flagship model.
Meta releases Llama 4
Meta ships Llama 4, its first mixture-of-experts flagship.
DeepSeek releases R1 open-weight reasoning model
DeepSeek R1 becomes a landmark open-weight reasoning release.
Learn
Claude 1: The First Model from Anthropic
Claude 1 is the initial public large language model from Anthropic, designed for strong context handling and assistant roles.
GPT-2: Second-Generation Generative Pretrained Transformer
GPT-2 is a second-generation language model known for its text generation capabilities, released with considerations for safety.
Open-source vs closed-source AI models
Weighing open-weight releases like Llama 4 and DeepSeek R1 against closed models like GPT-5 and Claude.