eicker@lemmy.world to Technology@lemmy.worldEnglish · 6 days agoNvidia just showed that the harness, not the AI model, is now the real hero: researchers got Claude Opus 5 to achieve a 100% score on the interactive reasoning benchmark ARC-AGI-3.techcrunch.comexternal-linkmessage-square4linkfedilinkarrow-up118arrow-down152
arrow-up1-34arrow-down1external-linkNvidia just showed that the harness, not the AI model, is now the real hero: researchers got Claude Opus 5 to achieve a 100% score on the interactive reasoning benchmark ARC-AGI-3.techcrunch.comeicker@lemmy.world to Technology@lemmy.worldEnglish · 6 days agomessage-square4linkfedilink
minus-squareDiurnambule@jlai.lulinkfedilinkEnglisharrow-up4arrow-down1·6 days agoAi = bad. Buy yeah, this open bar g doors for selfhosting LLMs. That mean frontier model are not that usefull
Ai = bad. Buy yeah, this open bar g doors for selfhosting LLMs. That mean frontier model are not that usefull