본문으로 건너뛰기
All news

xAI Launches Grok 4.5 Publicly — Opus-Class Performance at a Quarter of the Price

Summary: xAI released Grok 4.5 publicly on July 8, co-trained with Cursor and targeting real-world engineering tasks. Elon Musk positions it as Opus-class — but at a fraction of the cost.

Key Facts

  • Architecture: Built on xAI's 1.5T-parameter V9 foundation, trained across tens of thousands of NVIDIA GB300 GPUs
  • Pricing: $2/M input tokens, $6/M output tokens — 76% cheaper than Claude Opus 4.8 ($5/$25)
  • Access: Available in Grok Build, Cursor (all plans), and the xAI console API; EU availability expected mid-July
  • New integrations: Notion (knowledge management) and Convex (full-stack development) support announced July 9
  • A separate private-beta Grok 4.5 continues running internally at SpaceX and Tesla

Why It Matters

The frontier AI competition is shifting from raw benchmark performance toward price-efficiency. Grok 4.5's Cursor co-training signals a move toward real-world developer workflows over synthetic benchmarks. With Anthropic's Fable 5 export-control ban now lifted, all major labs are back in the public market simultaneously — intensifying the pricing race.

Read More

How this story unfolded

  1. Grok 4.5 Enters Private Beta at SpaceX and Tesla — 1.5T Parameters, Cursor-Trained
  2. Anthropic Releases Claude Sonnet 5: Near-Opus Performance at Sonnet Pricing
  3. xAI's Grok 4.3 Lands on Amazon Bedrock With 1M-Token Context and Native Video
  4. Alibaba's Qwen3-Coder-Next: 80B-Param MoE Coding Model With Only 3B Active — Scores 70.6% on SWE-Bench
LLM
— Large Language Model의 약자로, '거대 언어 모델'이라고 해요. ChatGPT, Claude 같은 AI가 바로 LLM이에요. 엄청나게 많은 텍스트를 학습해서 사람처럼 글을 쓰고 대화할 수 있어요.

뉴스레터 구독

무료 뉴스레터

매주 핵심 AI 소식, 한 번에 받기

쏟아지는 AI·LLM 뉴스 중 꼭 알아야 할 것만 골라 메일로 보내드려요. 뉴스레터 발송이 시작되면 구독자분들께 가장 먼저 보내드립니다.