A short moment to be proud of the work we do in refact.ai
We just updated the SWE-bench Multimodal leaderboard with new systems from @refact_ai, @allhands_ai and @TU_Muenchen. Congrats to all teams on pushing the state-of-the-art performance! SWE-bench Multimodal challenges AI systems to fix issues that are described using screenshots.
Inference cost will become the dominant operating expense for knowledge companies
Google has the best LLM model and the worst scaffolding around it in their @GeminiApp It can't understand its own format of short links which youtube provides when you click Share button. It encounters and error when you provide full format just like it expect it to see... 🤯…
AI/LLMs role is to augment humans not to replace them - from @karpathy Software 3.0 speech for YC School today. LLM is a software, a software with properties of savant with encyclopaedic knowledge and yet with many cognitive issues (hallucinations, amnesia, flawed…
Refact.ai got top score on the most popular Software Engineering Benchmark for AI Agents ( both SWE Verified and SWE Lite)
We tested Gemini 2.5 Pro with refact.ai on Polyglot Benchmark The model from Google got 82.2%, behind Claude 3.7 Sonnet’s 93.3% In the video I am providing the details how our model tests are different from aiders
Proud to participate in the upcoming @ainativedev conf ai-native-devcon-2025.heysummit.com/talks/use-of-f…
United States 趨勢
- 1. Halo 121K posts
- 2. PlayStation 56.6K posts
- 3. $BIEBER N/A
- 4. Xbox 70.3K posts
- 5. #WorldSeries 53.7K posts
- 6. #HitTheBuds 2,690 posts
- 7. Purdue 3,651 posts
- 8. Jasper Johnson N/A
- 9. End of 1 13.6K posts
- 10. #CashAppPools 1,658 posts
- 11. End 1Q N/A
- 12. Collin Chandler N/A
- 13. Cole Anthony N/A
- 14. End of the 1st 2,579 posts
- 15. Reagan 232K posts
- 16. Honeycutt N/A
- 17. Brandon Garrison N/A
- 18. Ashley 159K posts
- 19. Combat Evolved 5,147 posts
- 20. Rajah N/A
Something went wrong.
Something went wrong.