Conventional model routing sucks. It passes benchmarks but fails to write code you'd actually merge.... - 惯性聚合

推荐订阅源

博客园_首页

WordPress大学

美团技术团队

Google Developers Blog

大猫的无限游戏

Y Combinator Blog

博客园 - 【当耐特】

酷壳 – CoolShell

The Cloudflare Blog

The Blog of Author Tim Ferriss

奇客Solidot–传递最新科技情报

Google DeepMind News

MIT News - Artificial intelligence

Recent Announcements

Last Week in AI

宝玉的分享

The Register - Security

CTFtime.org: upcoming CTF events

cs.CL updates on arXiv.org

Hugging Face - Blog

Privacy & Cybersecurity Law Blog

Recorded Future

Help Net Security

News and Events Feed by Topic

博客园 - Franky

Proofpoint News Feed

LINUX DO - 热门话题

SegmentFault 最新的问题

The GitHub Blog

KPMG report finds enterprise disconnect between AI and its ROI | CIO

钛媒体：引领未来商业与生活新知

Google DeepMind News

有赞技术团队

Security Latest

阮一峰的网络日志

Java Code Geeks

Cognition(@cognition_labs)

Cognition (@cognition) on X Cognition (@cognition) on X Cognition (@cognition) on X Cognition (@cognition) on X Cognition (@cognition) on X Cognition (@cognition) on X SWE-1.7 is free for all paid users for the next month. We're also launching SWE-1.7 Lightning, serv... We trained SWE-1.7 in the Devin harness and taught it to self-compact on longer horizon tasks. As a... SWE-1.7 is available today in Devin across Web, Desktop, and CLI, running at 1000 tok/s. More detai... Our RL training spans four datacenters across three continents, combining our own GPUs across multip... SWE-1.7 was built on broad improvements in RL pipeline on top of a Kimi K2.7 base model. Our propri... Introducing SWE-1.7, the most capable model we’ve trained yet. It scores within a few points of the... We’ve made improvements to the FrontierCode methodology and are releasing FrontierCode 1.1 with clea... Research, design, and math are three fields that tend to attract obsessives. Few have had a bigger i... Read the announcement: https://t.co/M0U1jH4keS Our goal is to help security teams shift from reactive triage to proactively fixing what matters, cl... Building on Devin's new capabilities, today we're announcing the Security Vulnerability Remediation ... Try it in Devin! https://t.co/5s8NUGZXMS Claude Fable 5 is available in Devin. You can use Fable 5 in Devin Cloud’s Ultra agent, our smartes... We’re also publishing extensive documentation and technical materials about Agentic MapReduce, inclu... Security Swarm is a new pillar of Devin for Security: a suite of tools to help you find vulnerabilit... In testing, Devin Security Swarm found 36 of 50 real-world GHSA vulnerabilities at 30% lower cost pe... Introducing Devin Security Swarm A more cost effective and accurate way to find security vulnerabil... Through Aug 31, Sonnet 5 will use roughly 30% less quota than Sonnet 4.6 in Devin Desktop/CLI. After... Claude Sonnet 5 is now available in Devin Desktop and Devin CLI. Sonnet 5 pairs frontier-level codi... On FrontierCode (Extended), our benchmark for real-world engineering tasks that grades mergeability ... Watch Devin Fusion work: The sidekick model handles fetching information and skimming sources. The... For more research on how we built Devin Fusion, including examples, check out our blog: https://t.c... Devin Fusion is available today in Devin. Try it out: https://t.co/FcqbSxz7GB Devin Fusion uses a hybrid-model harness built around two ideas: First, a “sidekick” agent: a small... Second, Fusion routes dynamically mid-session. Engineering tasks often reveal their difficulty late... Most SOC2 audits take weeks of manual time. Read how @s16h_ and the @MetaviewAI team finished it in... Try Kimi K2.7 and GLM 5.2 for free in Devin Desktop and CLI Devin tests its work before you review the PR. You review and approve the test plan. Then get back... Read the docs for more details on testing: https://t.co/kOmHSo5hmn Every PR you open in Devin Review now gets a security review. Try it on your next one: https://t.c... Security review is now part of every Devin Review. Open a PR and Devin automatically finds the vuln... Each finding is ranked by severity, tagged with a CWE ID, and comes with a fix drafted as a merge-re... We’re removing access to Claude Fable 5 in Cognition products following Anthropic’s latest announcem... Last week we introduced the AI Productivity Guarantee: a $10,000,000 bet that Devin delivers enginee... Congratulations to @axliu42 on receiving the ACM Doctoral Dissertation Award for his work on learnin... /handoff is one of the most powerful features of Devin CLI: close your laptop, your agents keep work... @mercor_ai @Etched @AnthropicAI Apply here: https://t.co/qTo0gXQssQ June 19-20 · SF Two days left to apply for the Inference-Time Compute Hackathon, hosted by Cognition, @mercor_ai, @E... Read our blog post: https://t.co/gZkNRFqKEY You can try Claude Fable 5 as part of Devin Cloud’s Ultra agent. Devin Ultra is our smartest and mos... Try out Devin today! https://t.co/5s8NUH0vCq Claude Fable 5 is now available in Devin. Fable 5 earns the #1 spot on FrontierCode, our benchmark ... You can find full model results and technical implementation details on our blog: https://t.co/01vm... FrontierCode has three task sets: Extended (150 tasks), Main (100 tasks) and Diamond (50 tasks). SOT... Tasks in the dataset have a concise problem statement with large solutions that cut across multiple ... Introducing FrontierCode: a coding eval that raises the bar for difficulty & quality. Each task ... More than half of Cognition's first 50 employees were former founders. @ScottWu46 sat down with @b... Thank you to @lux_capital for hosting. Full video: https://t.co/FPTMbeRZoQ Here's the technical writeup on how the measurement works: https://t.co/lKouWoNx0q

Conventional model routing sucks. It passes benchmarks but fails to write code you'd actually merge....

Cognition (@ · 2026-06-29 · via Cognition(@cognition_labs)

此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。原文来自 — 版权归原作者所有。