Anthropic
Focus on AI safety, Constitutional AI, and large-context reasoning.
2025-05
Claude 4 Family
Thinking / Tools Specs: Major Version
Major version with extended thinking capabilities and advanced tool support.
2024-09
Safeguards Collaboration
Safety / Governance Specs: US/UK AISI Partnership
International collaboration to formalize AI safety evaluation standards.
2024-06
Claude 3.5 Sonnet
Multimodal Specs: Speed/Intellect
Mid-sized model outperforming previous flagship Opus at higher speeds.
2024-03
Claude 3 Family
Multimodal Specs: Haiku / Sonnet / Opus
Opus achieves SOTA on many benchmarks. tiered model system.
2023-07
Claude 2
LLM (Chat) Specs: 100k Context / Vision
Public release rival LLM with multimodal vision input and 100k context.
2023-03
Claude 1
LLM (Chat) Specs: Constitutional AI
First release focused on helpfulness and safety via RLAIF.