Social Wire
Page 2 of 33 · newest first.
@marcthecreatorr · Cognition
Gemini 4 Argon is now live, with access currently limited but expected to expand soon@marcthecreatorr · CLI
Gemini 4 Argon outperforms Opus 5.5 in 3D tasks, offering faster and more detailed results at a lower cost@notjazii Ambassador · Cognition
Gemini 4 Argon demonstrates strong performance in design and 3D tasks, including creating a Minecraft clone@leomarciano Ambassador · Cloud
Leo Marciano shares that his Devin /plan workflow now includes a dedicated debug step for every issue the agent creates, ensuring problems are investigated as they surface rather than deferred.@harshil1712 Ambassador · Cloud
Harshil Agrawal shares how Cloudflare's new Issues feature triggers Devin workflows automatically when production errors occur, launching a cloud agent to investigate and prepare fixes without manual triage.@notjazii Ambassador · CLI
Gemini 4 Argon outperforms Opus 5.5 in 3D tasks, offering similar results at a lower cost and faster speed@walden_yan · Cloud
Walden Yan shared how Cognition worked with Turbopuffer, Reducto, and LiteLLM to build the Pylon integration that lets engineers ship product instead of triaging support tickets.@sqs · Cognition
Devin users can now access their Claude plan in Amp, making it free and open to everyone, a highly requested feature.@adolandev Ambassador · Cloud
Developer @adolandev built an integration letting Nous Research Hermes drive Devin. The setup codes locally or hands off to Devin Cloud, where work runs on dedicated Linux, macOS, or Windows machines.@bryllim_ Ambassador · Cognition
The Cognition Devin Hackathon in Manila drew over 1,200 participants across 400+ teams, with 281 submissions—all organized in 11 days, per @bryllim_.@cognition · Cognition
Cognition's FrontierCode 1.1 benchmark is now live on Artificial Analysis. The benchmark, built with open-source maintainers, measures end-to-end code quality and whether pull requests are actually mergeable—the first to test if maintainers would merge the AI's work.@emergingbits · Cognition
A developer has successfully used DevinAI to automate marketing and ads, creating a 'mad factory' that streamlines processes and improves efficiency.@Fei2411 Ambassador · Cloud
Developer @Fei2411 released an open-source deepseek bot supporting custom models on macOS and Windows, with cross-platform testing done on Devin Cloud hosts.@frmdomi Ambassador · Cloud
Devin now has a dedicated Learning page to help users get started with Devin and its tools, notes @frmdomi.@jorgesancha · Cognition
A user's first impression of DevinAI is overwhelmingly positive, with a sense of excitement about the potential benefits.@jensenloke Ambassador · Cloud
Developer @jensenloke ran roughly 300 Devin Cloud sessions in a week with minimal usage impact, but 43 sessions on SWE 2.0 burned through 40% of a Cloud Max plan in one day—a reminder to plan capacity for the more capable models.@JustinGorya Ambassador · CLI
Discover how Opus 5.5 and Haiku 5.5 can enhance your workflow with native support in DevinCLI and Devin desktop via Fusion.@kr0der Ambassador · Cloud
Devin Cloud automatically sends feedback to Cognition when it detects the same error repeatedly, @kr0der notes—a smoother experience than manually submitting feedback in some other tools.@notjazii Ambassador · Cloud
Developer @notjazii tested Muse Wonder against Devin's SWE-2 with the same prompt at highest reasoning. Muse took 24 minutes and cost roughly $18, while SWE-2 completed in 15 minutes at no charge—and the quality surprised them.@heimlabs · Cognition
Devin's AI has a unique 'dreaming' pass, where it revisits and tidies memories as Git notes, once a day, showcasing its advanced capabilities.@pedrotabio · Cognition
A team is enthusiastically working on building and deploying Devin, anticipating a generational impact and significant advancements.@therceman Ambassador · Cloud
Developer @therceman chains Devin Mobile with ChatGPT's mobile app: Devin opens a pull request, ChatGPT reviews and commits changes, and Devin picks up the new commit automatically to continue work.@xishu_ai · Cognition
Cognition users are experiencing confusion with GitHub integration and unexpected charges despite connecting ChatGPT plans, highlighting a need for clarity and potentially a fix.@cognition · Cloud
Bring your own ChatGPT subscription to Devin. Connect a Go, Plus, or Pro plan and every GPT model call in Devin draws from your ChatGPT quota instead of Devin credits—useful if you already have a ChatGPT plan and want to maximize it across tools.