2026-08-07

2026-08-07에 감지된 AI 모델 가격·컨텍스트·종료일 변경 전체입니다.

변경모델내용
종료 예정openai/gpt-5.2-chat2026-08-10 종료 · D-3 남음
종료 예정openai/gpt-5.3-chat2026-08-10 종료 · D-3 남음
종료 예정z-ai/glm-4.52026-12-31 종료 · D-146 남음
가격 변경deepseek/deepseek-v4-flash출력 +58.7% · 캐시 읽기 +58.7% · 입력 +58.7%
가격 변경inclusionai/ling-3.0-flash출력 -71.4% · 캐시 읽기 -72.0% · 입력 -72.0%
가격 변경moonshotai/kimi-k2.6출력 +1.7% · 캐시 읽기 +1.7% · 입력 +1.7%
가격 변경qwen/qwen3-vl-235b-a22b-thinking출력 +1.3% · 입력 -59.2%
가격 변경qwen/qwen3.5-122b-a10b출력 +15.4% · 입력 +11.5%
가격 변경thinkingmachines/inkling-small입력 -10.0%
가격 변경z-ai/glm-5.2출력 -5.4% · 캐시 읽기 -5.4% · 입력 -5.4%
가격 변경~moonshotai/kimi-latest입력 -13.8%
당사자 발언Tibo @thsottiauxChatGPT 무료 사용자는 이제 GPT-5.6 Luna로 구동되는 무제한 텍스트 채팅을 이용할 수 있습니다 https://t.co/eZq1e1hTQX || Free users of ChatGPT now have unlimited text chats, powered by GPT-5.6 Luna https://t.co/eZq1e1hTQX
당사자 발언Tibo @thsottiauxGPT-5.6 Sol을 사용하는 Codex에게 가장 터무니없는 일들을 그냥 부탁해도, 그냥 해낸다. 나는 몇 주의 작업이 필요해 보이는 일들에 대해 그것과 5분 동안 쉬지 않고 이야기하고, 냉장고에서 뭔가를 꺼내러 일어나고, 개를 쓰다듬고, 돌아오면 ... 끝나 있다. || You can just ask Codex with GPT-5.6 Sol the wildest things and it will just do it. I talk to it for 5 minutes straight with things that just seem to require weeks of work, get up to get something in t
당사자 발언OpenAI @OpenAI당사의 출시 예정 모델 중 하나인 Astra를 평가한 후, 당사는 Preparedness Framework에 따라 이를 사이버 보안 분야의 첫 번째 "중대" 모델로 취급하고 있습니다. 이는 당사가 계획해 온 시나리오이며, Astra의 추가 개발을 보장하기 위해 추가 통제 조치를 마련하고 있습니다 || After evaluating one of our upcoming models, Astra, we're treating it as our first "critical" model for cybersecurity under our Preparedness Framework. This is a scenario we've planned for, and we're
RELEASEv0.121.0
RELEASEv2.1.224
RELEASE0.147.0
RELEASEb10321
ANNOUNCEMENTSet pull request limits at the organization level
ANNOUNCEMENT“Relates to” issue relationship in public preview and multi-select fields is generally available
ANNOUNCEMENTCopilot usage metrics API adds agent app activity
ANNOUNCEMENTGitHub Code Quality no longer adds Copilot as a reviewer
ANNOUNCEMENTSecret scanning coverage updates
ANNOUNCEMENTA guide to slash commands in the GitHub Copilot app
ANNOUNCEMENTTutorMoments: Do AI tutors know when to help and when to hold back?
ANNOUNCEMENTHow HSP GRUPPE builds AI capabilities for tax advisory
ANNOUNCEMENTResponding to the next frontier of critical cyber capabilities
파라미터 변경anthropic/claude-opus-4.1삭제 max_completion_tokens, response_format, structured_outputs

이전 2026-08-06