広告
主要AIモデルのリリース・アップデート履歴を時系列で表示
Claude 4 family released. Opus 4 achieves highest coding accuracy (SWE-Bench 80.9%).
1M token context with enhanced multimodal capabilities and Deep Research.
GPT-4.1 released with 1M token context support and improved instruction following.
Sora video generation integrated into ChatGPT Plus.
Sonnet 4.5 released as new default model with improved speed and quality.
Grok 3 released with 2M token context — industry's largest.
Open source LLaMA 4 released under permissive license.
Ultra-fast and cost-effective model with 1M token context.
Reasoning-focused models with chain-of-thought capabilities.
Open source reasoning model rivaling GPT-4o on math benchmarks (~90% AIME).
Gemini 2.0 launched with native multimodal capabilities.
Fast, affordable model for high-throughput tasks.
Major upgrade with computer use capability and improved coding.
First reasoning model with chain-of-thought, targeting math and science.
Balanced performance at lower cost, outperforming Claude 3 Opus.
Omni model with native voice, vision, and text in a single model.
Haiku, Sonnet, Opus launched with breakthrough performance.
Revolutionary 1M token context window.
Open source mixture-of-experts model achieving GPT-3.5 level.
128K context, faster and cheaper GPT-4 variant.