2026-07-29

2026-07-29에 감지된 AI 모델 가격·컨텍스트·종료일 변경 전체입니다.

변경모델내용
가격 변경z-ai/glm-5.2출력 -4.8% · 캐시 읽기 -4.8% · 입력 -4.8%
가격 변경qwen/qwen3-coder-next출력 -11.1% · 캐시 읽기 +94.4% · 입력 -33.3%
가격 변경z-ai/glm-5.2출력 -1.9% · 캐시 읽기 -1.9% · 입력 -1.9%
당사자 발언Sam Altman @samachatgpt work는 놀랍다. 그리고 "work"라는 말로는 그 진가를 다 표현하지 못한다. 휴대폰으로 이렇게 보냈다: "내 모든 채팅 기록을 활용해서 친구 8명과 함께 갈 긴 주말 여행 아이디어를 찾아내고, 가장 좋은 세 가지 선택지를 계획하고, 우리 9명이 무엇을 하고 싶은지 조율할 수 있는 풀스택 사이트를 만들어 줘 || chatgpt work is remarkable, and "work" undersells it. from my phone i sent: "use all my chat history to figure out ideas for a long weekend trip with 8 friends, plan the best three options, make a ful
당사자 발언Elon Musk @elonmuskGrok Voice가 이제 에이전트 성능에서 #1입니다 https://t.co/1MxiX1MPn1 || Grok Voice is now #1 in agentic performance https://t.co/1MxiX1MPn1
당사자 발언Dario Amodei @DarioAmodei투명성에 더해, 이제 나는 프런티어 모델이 사이버, 생물학 및 자율성 위험에 대해 의무적인 제3자 테스트를 받아야 하며, 치명적인 위험을 초래하는 모델의 배포를 차단하거나 철회할 권한도 있어야 한다고 믿는다. || In addition to transparency, I now believe frontier models should face mandatory third-party testing for cyber, bio, and autonomy risks—with the power to block or revoke deployment of models that pose
당사자 발언Tibo @thsottiaux알고 보니 GPT-5.6 Sol은 실제로 ARC-AGI-3에서 SoTA다. 설정 두 가지만 바꾸면 됐다. 우리의 표준 compaction 구현을 활용해 여러 context window에 걸쳐 추론하고 작업하도록 허용하기만 하면 된다. https://t.co/36rofGztdS https://t.co/I5C2xcsm9y || Turns out GPT-5.6 Sol is actually SoTA on ARC-AGI-3. Just took two setting changes. You just have to allow it to reason and work over multiple context windows with the help of our canonical compaction
당사자 발언Tibo @thsottiaux효율성! 두 단계로 a) 훌륭한 모델을 훈련한다 b) 훌륭한 모델을 사용해 자체 인프라, 추론 스택, 커널 등을 포함한 모든 것을 개선한다 https://t.co/kGtmX3LAow https://t.co/qrfNtdkYYA || Efficiency! In two steps a) Train fantastic model b) Use fantastic model to make everything better, including its own infrastructure, inference stack, kernels, etc, etc https://t.co/kGtmX3LAow https:/
당사자 발언OpenAI @OpenAI스택 전반의 이러한 최적화가 복합적으로 작용해 비용-지능 곡선의 모든 지점에서 가장 뛰어난 성능의 모델을 사용할 수 있게 합니다. https://t.co/tcv8oo0z7G || These optimizations across our stack compound to unlock the most performant models at every point in the cost-intelligence curve. https://t.co/tcv8oo0z7G
당사자 발언OpenAI @OpenAI배포 후에는 GPT-5.6 Sol을 활용해 자체 실행 효율을 높임으로써 효율성의 최전선을 확장했습니다. 결과: - 프로덕션 GPU 커널 개선으로 서빙 비용 20% 절감. - 개선된 추측 디코딩으로 토큰 생성 효율 15%+ 향상. || After deployment, we applied GPT-5.6 Sol to advance the frontier of efficiency by making itself more efficient to run. The results: - 20% lower serving costs from production GPU kernel improvements. -
당사자 발언OpenAI @OpenAI참가자는 GPT-5.6 모델 제품군을 포함한 당사의 프런티어 모델을 이용할 수 있습니다. 각 workspace에는 비즈니스급 개인정보 보호 및 보안 기능이 포함됩니다. 연구자 데이터는 기본적으로 당사 모델의 훈련에 사용되지 않으며, 참가자는 최대 4명까지 초대할 수 || Participants will receive access to our frontier models, including our GPT-5.6 family of models. Each workspace includes business-grade privacy and security protections. Researcher data is not used to
당사자 발언OpenAI @OpenAI과학자, 수학자, 엔지니어에게 당사의 프런티어 모델을 무료로 제공합니다. 연구자 10,000명부터 시작해 2027년까지 100,000명으로 확대합니다. ChatGPT for Academic Researchers는 여러 학문 분야의 발견을 가속하도록 설계되었습니다. https://t.co/nrHn1no0wQ || We’re giving scientists, mathematicians, and engineers free access to our frontier models—starting with 10,000 researchers and expanding to 100,000 through 2027. ChatGPT for Academic Researchers is bu
당사자 발언Anthropic @AnthropicAI또한 ETH Zurich, Tel Aviv University, University of Haifa의 학자들과 협력해 LLM의 암호해독 능력을 연구하기 위한 벤치마크인 CryptanalysisBench를 구축했습니다. https://t.co/OEYhGjH5Lj || We also worked with academics at ETH Zurich, Tel Aviv University, and the University of Haifa to build CryptanalysisBench, a benchmark for studying LLMs’ cryptanalysis abilities. https://t.co/OEYhGjH5
당사자 발언Anthropic @AnthropicAI두 공격의 전체 기술적 세부 사항은 새로운 논문에 담았습니다: HAWK 관련: https://t.co/JtLzetLOtU AES 관련: https://t.co/7pozPjc4o8 그리고 AES와 관련된 모델의 사고 과정: https://t.co/Cg1tfTJhvi || Full technical details of both attacks are provided in our new papers: On HAWK: https://t.co/JtLzetLOtU On AES: https://t.co/7pozPjc4o8 And the associated model chain-of-thought for AES: https://t.co/
당사자 발언Anthropic @AnthropicAI그럼에도 두 결과 모두 프런티어 AI 모델이 전문가 수준의 암호학 연구를 수행할 수 있음을 보여줍니다. 이는 중요한 방어적 활용 가능성이 있습니다. 우리의 온라인 활동을 안전하게 지키는 알고리즘을 테스트하고, 궁극적으로 디지털 시스템을 더 안전하게 만드는 데 도움을 줄 수 있습니다. || Still, both results show that frontier AI models are capable of doing expert-level cryptography research. This has important defensive applications—testing the algorithms that keep our online activity
당사자 발언Google AI Developers @googleaidevsGemini 3.5 Flash-Lite가 대규모의 반복적인 시각 작업을 어떻게 처리하는지 확인하고 싶었습니다. 이 데모는 1M+개의 카탈로그 이미지에 모델을 실행하여, 대규모 workflow에 필요한 낮은 지연 시간과 토큰 효율성을 유지하면서 원시 특징을 깔끔하고 구조화된 데이터로 추출합니다. https://t.co/8XpwhfRplL || We wanted to see how Gemini 3.5 Flash-Lite handles massive, repetitive visual tasks. This demo runs the model across 1M+ catalog images, extracting raw features into clean, structured data with the lo
당사자 발언Google AI Developers @googleaidevs👾Gemini 3.6 Flash로 구축한 멀티 에이전트 시스템이 플레이 가능한 게임 디자인을 실시간으로 반복 개선하는 모습을 확인해 보세요. 빠르게 전개되는 환경에서 속도와 추론의 균형을 맞추며, 실시간 플레이어 행동을 바탕으로 퍼즐 도전 과제를 설계, 구성, 검증하는 데 Gemini 3.5 Flash-Lite를 사용합니다. https://t.co/uQYbIvTcaW || 👾Watch a multi-agent system built with Gemini 3.6 Flash iterate on playable game design in real time. It uses Gemini 3.5 Flash-Lite to design, construct, and verify puzzle challenges based on live pla
파라미터 변경mistralai/ministral-8b-2512삭제 logprobs, repetition_penalty, top_logprobs
파라미터 변경qwen/qwen3-30b-a3b삭제 logprobs, structured_outputs, top_logprobs
파라미터 변경qwen/qwen3-30b-a3b-instruct-2507삭제 min_p

다음 2026-07-30