A HISTORY, ONE RELEASE AT A TIME
Meet the models.
What they could do. What changed. Why it mattered.
Claude Haiku 5.5
Fast assistance for frequent tasks
Gemini 4 Argon
A new frontier model with restricted access
GPT-6.1 Sol
Near Astra performance at a lower price
Claude Sonnet 5.5
Faster everyday coding and documents
GPT-6 Sol and Luna
GPT-6 at different cost levels
Claude Opus 5.5
Higher capability at lower operating cost
DeepSeek V4.1 Flash
Native visual understanding
GPT-6 Astra
A new generation for complex work
Gemini 3.8 Flash
Stronger reasoning in the Flash tier
Claude Mythos 5.1
Different safeguards for approved researchers
Claude Fable 5.1
Stronger research and professional work
DeepSeek V4 Pro
V4 Pro reaches general availability
Gemini 3.7 Flash
Stronger everyday agents
DeepSeek V4 Flash
Updated Flash agents
Claude Opus 5
Fifth generation Opus
GPT-5.6 family
More capability for different budgets
Claude Sonnet 5
Fifth generation Sonnet
Claude Mythos 5
Specialist research with restricted access
Claude Fable 5
A new tier for demanding work
Claude Opus 4.8
More consistent longer tasks
Gemini 3.5 Flash
Agents built to take action
DeepSeek V4 preview
A new long context family
GPT-5.5
Carry a larger task through to completion
Claude Opus 4.7
Coding, vision and complex tasks
GPT-5.4
Reasoning and coding for professional work
Gemini 3.1 Pro
Deeper reasoning for complex questions
Claude Sonnet 4.6
Broader computer and office work
GPT-5.3 Codex Spark
Coding with very low delay
Claude Opus 4.6
More reliable complex work
GPT-5.3 Codex
Coding agents you can steer while they work
Gemini 3 Flash
Reasoning at interactive speed
GPT-5.2
Documents, spreadsheets and complex projects
DeepSeek V3.2
Updated reasoning modes
Claude Opus 4.5
More capable coding and agents
Gemini 3 Pro
Reasoning across text and images
GPT-5.1 API
Adjust reasoning to the task
Claude Haiku 4.5
Capable coding in a smaller model
Claude Sonnet 4.5
Longer coding and computer tasks
DeepSeek V3.1
Thinking and ordinary answers
GPT-5
Fast answers and deeper reasoning
Claude Opus 4.1
Stronger software agents
Claude 4 (Opus 4 and Sonnet 4)
Longer coding workflows
Llama 4 Scout and Maverick
Open weight multimodal models
Gemini 2.5 Pro
Reasoning across long and visual inputs
Claude 3.7 Sonnet
Reasoning and coding in one model
DeepSeek R1
Reasoning with downloadable weights
OpenAI o1 (preview)
More computation devoted to reasoning
Llama 3.1 405B
A larger downloadable flagship
Claude 3.5 Sonnet
A stronger everyday coding assistant
GPT-4o
Text, vision and audio come together
Llama 3
Downloadable models become more useful
Claude 3 Opus
Stronger analysis with visual inputs
Gemini 1.5 Pro
Much longer inputs become possible
Gemini 1.0
A new multimodal model family
Llama 2
More accessible model weights
Claude 2
Longer documents enter the conversation
Claude 1
A new conversational assistant
GPT-4
A step forward in reasoning and images
ChatGPT (GPT-3.5)
Conversational AI reaches a wider audience
GPT-3
One model, many language tasks
GPT-2
A glimpse of fluent text generation
No models match that search. Try a provider or a different capability.