2026-08-05

2026-08-05에 감지된 AI 모델 가격·컨텍스트·종료일 변경 전체입니다.

변경모델내용
종료 예정openai/gpt-5.2-chat2026-08-10 종료 · D-5 남음
종료 예정openai/gpt-5.3-chat2026-08-10 종료 · D-5 남음
종료 예정z-ai/glm-4.52026-12-31 종료 · D-148 남음
가격 변경minimax/minimax-m2.5입력 +46.7%
가격 변경moonshotai/kimi-k2.7-code입력 -4.1%
가격 변경qwen/qwen3-vl-235b-a22b-thinking출력 +1.3% · 입력 -59.2%
RELEASE0.146.1
당사자 발언Elon Musk @elonmuskGrok Build https://t.co/H3SoRccsN3 || Grok Build https://t.co/H3SoRccsN3
당사자 발언OpenAI @OpenAI저희는 독립 평가 파트너들이 수행한 외부 사이버 평가 중에 발생한 두 건의 새로운 사건을 상세히 설명합니다. 저희는 무슨 일이 있었는지, 해당 활동이 어떻게 통제되었는지, 그리고 제3자 테스트에 대한 접근 방식을 강화하기 위해 평가자들과 어떻게 협력하고 있는지를 개괄합니다. || We're detailing two new incidents that occurred during external cyber evaluations conducted by independent evaluation partners. We outline what happened, how the activity was contained, and how we’re
당사자 발언Anthropic @AnthropicAI영국의 @AISecurityInst (AISI)는 Anthropic의 Claude Mythos 5와 OpenAI의 GPT-5.6 Sol에 대해 최근 실시한 사이버 보안 평가에 관한 보고서를 발표했습니다. 해당 모델들은 과제를 완료하려고 시도했으며, 그 환경에서는 모델들의 평소 안전장치가 제거되었고 모델들은 의도적으로 || The UK’s @AISecurityInst (AISI) has published a report on their recent cybersecurity evaluation of Anthropic’s Claude Mythos 5 and OpenAI’s GPT-5.6 Sol. The models attempted to complete an assignment
OUTAGEElevated errors across many models
OUTAGEDegraded performance of multiple models
OUTAGEDegraded performance for Claude Opus 5
RELEASEv2.1.222
RELEASEb10285
ANNOUNCEMENTCustomize code scanning default setup at scale
ANNOUNCEMENTRetiring the Copilot Billing Preview app
OUTAGESome Copilot Cloud Agent jobs not starting
ANNOUNCEMENTIntroducing the OpenAI Economic Research Exchange
ANNOUNCEMENTThird-party cyber evaluations involving OpenAI models
OUTAGEElevated errors with image generation
OUTAGEElevated errors with image generation
OUTAGEPartial System Degradation
컨텍스트 변경upstage/solar-pro-3128000 -> 131072
파라미터 변경upstage/solar-pro-3추가 frequency_penalty, presence_penalty, top_p

이전 2026-08-04