2026-08-14에 감지된 AI 모델 가격·컨텍스트·종료일 변경 전체입니다.
| 변경 | 모델 | 내용 |
|---|---|---|
| 가격 변경 | google/gemini-3.7-flash | web_search +100.0% |
| 가격 변경 | google/gemini-3.7-flash:batch | web_search +100.0% |
| 가격 변경 | moonshotai/kimi-k2.6 | 출력 -41.0% · 캐시 읽기 -41.0% · 입력 -41.0% |
| 가격 변경 | openai/gpt-5.6-luna | web_search +100.0% |
| 가격 변경 | openai/gpt-5.6-luna-pro | web_search +100.0% |
| 가격 변경 | openai/gpt-5.6-terra | web_search +100.0% |
| 가격 변경 | openai/gpt-5.6-terra-pro | web_search +100.0% |
| 가격 변경 | qwen/qwen3.5-35b-a3b | 출력 +44.0% · 캐시 읽기 -10.0% · 입력 -10.0% |
| 가격 변경 | z-ai/glm-4.6 | 출력 +10.0% · 캐시 읽기 +10.0% · 입력 +10.0% |
| 가격 변경 | z-ai/glm-5.2 | 출력 +88.9% · 캐시 읽기 +133.9% · 입력 +88.9% |
| 가격 변경 | ~deepseek/deepseek-v4-flash-latest | 출력 -36.7% · 캐시 읽기 -36.7% · 입력 -0.2% |
| 가격 변경 | ~google/gemini-flash-latest | web_search +100.0% |
| 가격 변경 | deepseek/deepseek-v4-flash-0731 | 출력 +55.6% · 캐시 읽기 +75.0% · 입력 +75.0% |
| 가격 변경 | google/gemini-3.6-flash | audio -50.0% · 출력 -50.0% · image -50.0% · input_audio_cache -50.0% · 캐시 읽기 -50.0% · 캐시 쓰기 -50.0% · internal_reasoning -50.0% · 입력 -50.0% |
| 가격 변경 | google/gemini-3.6-flash:batch | audio -50.0% · 출력 -50.0% · image -50.0% · input_audio_cache -50.0% · 캐시 읽기 -50.0% · 캐시 쓰기 -50.0% · internal_reasoning -50.0% · 입력 -50.0% |
| 가격 변경 | meta-llama/llama-4-maverick | 출력 +14.9% |
| 가격 변경 | moonshotai/kimi-k2.7-code | 출력 +2.9% · 입력 +6.0% |
| 가격 변경 | qwen/qwen3-vl-30b-a3b-instruct | 출력 -13.3% · 입력 -13.3% |
| 가격 변경 | qwen/qwen3.5-397b-a17b | 출력 +20.0% · 캐시 읽기 +566.7% · 입력 +11.1% |
| 가격 변경 | z-ai/glm-5.2 | 출력 -37.1% · 캐시 읽기 -5.5% · 입력 +26.0% |
| 가격 변경 | ~google/gemini-flash-latest | audio -75.0% · 출력 -75.0% · image -75.0% · input_audio_cache -75.0% · 캐시 읽기 -75.0% · 캐시 쓰기 -75.0% · internal_reasoning -75.0% · 입력 -75.0% · web_search -50.0% |
| 신규 모델 | google/gemini-3.7-flash | 컨텍스트 1,048,576 |
| 신규 모델 | google/gemini-3.7-flash:batch | 컨텍스트 1,048,576 |
| 모델 삭제 | inclusionai/ling-3.0-tiny:free | |
| 당사자 발언 | Sam Altman @sama | 여러분의 시스템을 방어하는 데 도움이 되도록 저희 모델 사용을 고려해 주세요 https://t.co/oqs3qfWmtN || please consider using our models to help defend your systems https://t.co/oqs3qfWmtN |
| 당사자 발언 | Tibo @thsottiaux | 하루가 지난 뒤 ChatGPT에게 당신의 컴퓨터 사용을 신랄하게 놀려 달라고 해보세요. Computer History 플러그인을 설치하고, 평소처럼 컴퓨터를 사용한 다음 당신의 행동을 신랄하게 놀려 달라고 정중히 요청하세요. 그건 ... 통찰력이 있어요 👀 ===== - 당신은 Slack을 사용하지 않아요. Slack이 당신을 사용해요. Slack은 기록된 활동의 48%를 차지했어요. 당신의 Mac https://t.co/fH || Ask ChatGPT to roast your computer usage after a day. Install the Computer History plugin, do things as usual and kindly ask it to roast your behavior. It's ... insightful 👀 ===== - You don’t use Slac |
| 당사자 발언 | Tibo @thsottiaux | ChatGPT 안에서 바로 Google docs, sheets and slides로 작업하세요. 이것은 제가 문서를 작성하고, 브레인스토밍하거나, 내용을 교정하는 방식을 바꿔 놓았습니다. 그냥 열고 나서 채팅하거나 말로 변경 사항을 하나씩 진행하면, 모든 일이 바로 그 흐름 속에서 일어납니다. https://t.co/iiz1eKyp46 || Work with Google docs, sheets and slides right inside ChatGPT. This has changed how I write documents, brainstorm or proofread things. I just open it and then chat or talk my way through changes and i |
| 당사자 발언 | OpenAI @OpenAI | Computer History를 사용하려면 Mac용 ChatGPT 데스크톱 앱의 Settings → Integrations에서 옵트인하세요. 현재 Pro, Business, Enterprise 사용자에게 전 세계적으로 출시되고 있으며, EEA, UK, 스위스에서는 앞으로 몇 주 안에 이용할 수 있게 됩니다. || To use Computer History, opt in under Settings → Integrations in the ChatGPT desktop app on Mac. Rolling out globally now to Pro, Business, and Enterprise users, with access in the EEA, UK, and Switze |
| 당사자 발언 | OpenAI @OpenAI | Computer History는 토큰 사용량을 줄이고 더 많은 개인정보 보호 제어 기능을 제공하는 Chronicle 연구 프리뷰를 기반으로 합니다. 새로운 타임라인 보기를 통해 작업을 되돌아보고 자주 하는 작업을 바탕으로 스킬을 만들 수 있습니다. 그곳이나 메뉴 막대에서 다음을 할 수 있습니다: - 모두 지우거나 || Computer History builds on the Chronicle research preview with reduced token usage and more privacy controls. A new timeline view gives you the ability to look back on your work and build skills from |
| 당사자 발언 | OpenAI @OpenAI | ChatGPT는 이제 컴퓨터의 앱과 웹사이트 전반에서 사용자의 활동을 기억할 수 있습니다. 데스크톱 앱에서 Computer History를 사용하면 향후 상호작용이 더 개인화된 느낌을 주며 설명이 덜 필요합니다. https://t.co/WHZPxPp31R || ChatGPT can now remember your activity across the apps and websites on your computer. With Computer History in the desktop app, future interactions feel more personalized and require less explanation. |
| 당사자 발언 | OpenAI @OpenAI | @Cerebras로 구동되는 Ultrafast는 최대 750 tokens per second를 생성해, 매초가 중요한 제품과 워크플로에 저희의 가장 지능적인 모델을 제공합니다. Ultrafast는 더 빠른 프런티어 지능이 측정 가능한 이점을 만들어 내는 기업을 위해 설계되었으며, 여기에는 || Powered by @Cerebras, Ultrafast generates up to 750 tokens per second, bringing our most intelligent model to products and workflows where every second counts. Ultrafast is designed for businesses whe |
| 당사자 발언 | Anthropic @AnthropicAI | Responsible Scaling Policy의 일환으로, 저희는 정기적인 Risk Reports를 발행합니다. 이 보고서들은 저희 시스템의 위험과 이를 해결할 준비가 얼마나 되어 있는지에 대한 자세한 정보를 공유합니다. 두 번째 Risk Report를 이제 확인할 수 있습니다: https://t.co/NgWnDmXZD3 || As part of our Responsible Scaling Policy, we publish regular Risk Reports. These share detailed information on the risks of our systems and how prepared we are to address them. Our second Risk Report |
| 당사자 발언 | Anthropic @AnthropicAI | 저희는 미공개 연구 버전의 Claude에게 리만 가설을 한번 풀어 보도록 요청했습니다. 이를 풀지는 못했지만, 관련 문제에서는 진전을 이루었습니다. 가설을 만족하는 리만 제타 함수의 영점 비율에 대한 하한을 다음 수치에서 높였습니다 || We asked an unreleased research version of Claude to take a stab at the Riemann hypothesis. It didn’t solve it, but it did make strides on a related problem: it increased the lower bound for the fract |
| 당사자 발언 | Google AI Developers @googleaidevs | 아이디어를 기획하고 앱을 만들었습니다. 이제 성장할 시간입니다! 📈 August 17까지 진행되는 Build with Gemini @XPrize Hackathon을 위해 @FlowbyGoogle과 @PomelliByGoogle 같은 도구를 사용하여 제품을 마케팅하고 창의적인 캠페인을 전개하는 방법을 알아보세요. https://t.co/BaJftVSaA0 || You’ve planned your idea and built your app. Now it’s time to grow! 📈 Learn how you can market your product and deploy creative campaigns using tools like @FlowbyGoogle and @PomelliByGoogle for the Bu |
| 당사자 발언 | Google AI Developers @googleaidevs | 하나의 사양. 다섯 개의 네이티브 앱. 보일러플레이트 제로. @antigravity의 Gemini 3.7 Flash가 단일 아키텍처 사양을 받아 5개의 모바일 프레임워크 전반에서 네이티브이며 프로덕션에 사용할 준비가 된 코드를 생성하는 모습을 확인하세요: Flutter, SwiftUI, Jetpack Compose, React Native, NativeScript. https://t.co/1nHKmuSMdW || One spec. Five native apps. Zero boilerplate. Watch how Gemini 3.7 Flash in @antigravity takes a single architecture specification and generates native, production-ready code across 5 mobile framework |
| RELEASE | langchain-core==1.5.5 | |
| RELEASE | b10433 | |
| RELEASE | v0.32.12 | |
| ANNOUNCEMENT | GitHub Copilot weekly releases — August 10 | |
| ANNOUNCEMENT | Grok 4.6 is now available in GitHub Copilot | |
| ANNOUNCEMENT | How to bring your software delivery workflow into GitHub with agent apps | |
| ANNOUNCEMENT | State of Open Models: Summer 2026 Observations | |
| 파라미터 변경 | deepseek/deepseek-v4-flash | 추가 max_completion_tokens |
| 파라미터 변경 | deepseek/deepseek-v4-pro | 추가 max_completion_tokens |
| 컨텍스트 변경 | qwen/qwen3.8-2.4t-a95b | 1010000 -> 1048576 |
| OUTAGE | Elevated errors for Claude Mythos 5, Claude Fable 5, and Claude Sonnet 5 | |
| OUTAGE | Issues reaching status.claude.com | |
| OUTAGE | Degraded performance for multiple models | |
| OUTAGE | Service disruption on Claude Code on the web and related services | |
| ANNOUNCEMENT | Introducing Gemini 3.7 Flash | |
| ANNOUNCEMENT | Putting sign language AI into users’ hands | |
| RELEASE | v0.122.0 | |
| RELEASE | v2.1.232 | |
| RELEASE | langchain-core==1.5.4 | |
| RELEASE | b10428 | |
| RELEASE | v0.32.11 | |
| RELEASE | v3.0.0 | |
| RELEASE | v2.18.1 | |
| RELEASE | Release: v5.15.0 | |
| RELEASE | v0.27.1 | |
| ANNOUNCEMENT | Automatically migrate branch protection rules to repository rulesets | |
| ANNOUNCEMENT | Copilot memory and Ollama in GitHub Copilot for JetBrains | |
| ANNOUNCEMENT | GitHub Enterprise Server 3.22 release candidate | |
| ANNOUNCEMENT | MAI-Code-1.1-Flash available in GitHub Copilot | |
| ANNOUNCEMENT | Upcoming deprecation of MAI-Code-1-Flash | |
| ANNOUNCEMENT | Agent Plugins 1.0 in VS Code, Copilot CLI, and the Copilot app | |
| ANNOUNCEMENT | Rule insights for organizations in public preview | |
| ANNOUNCEMENT | Block users from comments in personal repositories | |
| ANNOUNCEMENT | Gemini 3.7 Flash is now available in GitHub Copilot | |
| ANNOUNCEMENT | License data quality improvements | |
| ANNOUNCEMENT | Using the GitHub Copilot SDK for Java | |
| ANNOUNCEMENT | From coder to orchestrator: How agents shift the role of a developer | |
| ANNOUNCEMENT | Your contributors are AI-first now. Is your project? | |
| ANNOUNCEMENT | What 50 open source projects taught us about security in the AI era | |
| ANNOUNCEMENT | Write your first prompt with the GitHub Copilot app | |
| ANNOUNCEMENT | GitHub availability report: July 2026 | |
| ANNOUNCEMENT | Your guide to GitHub Universe 2026 is here: The schedule just launched! | |
| OUTAGE | Disruption with Copilot for access to some models | |
| OUTAGE | Errors with the Fable 5 Model in Copilot | |
| OUTAGE | Incident with Pull Requests and Issues | |
| OUTAGE | Incident with Webhooks | |
| OUTAGE | Disruption with Login and Release Asset downloads | |
| OUTAGE | Disruption with GHEC Team Sync | |
| OUTAGE | Disruption with creation of fine grained personal access tokens | |
| OUTAGE | Incident with GraphQL API Requests | |
| ANNOUNCEMENT | LFM2.5-VL-3B for Better and Faster Vision Capabilities for the Edge | |
| ANNOUNCEMENT | Making Knowledge Distillation Cheap Enough to Run at Scale | |
| ANNOUNCEMENT | Introducing OlmoEarth embeddings: Custom embedding exports from OlmoEarth Studio for downstream analysis | |
| ANNOUNCEMENT | Record, train, and deploy from one place with Strands Agents, LeRobot, and Hugging Face Storage Buckets | |
| ANNOUNCEMENT | What We Learned by Reproducing 2,200 papers from ICML | |
| ANNOUNCEMENT | Meta is back with Muse Glimmer: local, agentic, multimodal, and open source | |
| ANNOUNCEMENT | Build Low-Latency Multilingual Voice Agents: Open Weights & Full Deployment Control with NVIDIA Magpie TTS | |
| ANNOUNCEMENT | How business operations teams use ChatGPT Work | |
| ANNOUNCEMENT | How data science teams use ChatGPT Work | |
| ANNOUNCEMENT | How sales teams use ChatGPT Work | |
| ANNOUNCEMENT | The builder’s guide to GPT‑5.6 | |
| ANNOUNCEMENT | What building an AI-native finance function taught me | |
| ANNOUNCEMENT | OpenAI appoints Dali Rajic as Chief Revenue Officer | |
| ANNOUNCEMENT | Daybreak models are now available on AWS | |
| ANNOUNCEMENT | Expanding Daybreak as the Cyber Defense Window Narrows | |
| ANNOUNCEMENT | From assistance to execution: How enterprises put AI to work | |
| ANNOUNCEMENT | Model ML completes finance work more efficiently with GPT-5.6 Sol | |
| ANNOUNCEMENT | Premium seats are coming to ChatGPT Business | |
| ANNOUNCEMENT | Previewing Ultrafast mode: GPT-5.6 Sol at up to 14X the speed | |
| ANNOUNCEMENT | Putting frontier cyber models in more trusted hands | |
| ANNOUNCEMENT | OpenAI’s letter to Governor Abbott on responsible AI infrastructure in Texas | |
| ANNOUNCEMENT | How RingCentral builds AI-native work from engineering to ops | |
| ANNOUNCEMENT | Virgin Atlantic sharpens customer journeys with ChatGPT Work | |
| ANNOUNCEMENT | How Zapier transformed core marketing processes with ChatGPT Work | |
| OUTAGE | Increased errors for some ChatGPT users | |
| OUTAGE | Increased error rates | |
| OUTAGE | Increased error rates | |
| OUTAGE | Elevated errors affecting ChatGPT Go conversations | |
| OUTAGE | Users unable to login to ads manager | |
| OUTAGE | Increase in errors on API, Codex and Work Mode | |
| OUTAGE | Error while creating custom RBAC roles for Enterprise users | |
| OUTAGE | Elevated errors in ChatGPT conversations for Free users | |
| 컨텍스트 변경 | anthracite-org/magnum-v4-72b | 16384 -> 32768 |
| 컨텍스트 변경 | deepseek/deepseek-r1 | 163840 -> 64000 |
| 파라미터 변경 | deepseek/deepseek-r1 | 삭제 max_completion_tokens |
| 파라미터 변경 | deepseek/deepseek-v4-pro-0813 | 추가 logit_bias, min_p, repetition_penalty, seed, structured_outputs, top_k |
| 파라미터 변경 | kwaipilot/kat-coder-air-v2.5 | 삭제 frequency_penalty |