Trending Discussions
LocoMod (⬆270): Keep in mind that this is NOT the agentic coding index from Artificial Analysis. It’s just a poorly made knockoff benchmark site deliberately designed to look like AA’s official one. They did this intentionally, so that site can’t be trusted at all. Just wait for the official benchmarks from AA. fvancesco (⬆64): As if AA is any more trustworthy. MerePotato (⬆27): They’re pretty solid, provided you look at the individual benchmark scores to get the full picture rather than just relying on the aggregate overall score. Their model comparison tool is really useful. sloptimizer (⬆17): You say that like Artificial Analysis is actually reliable. Available_Age8480 (⬆18): These benchmarks are even more absurd than the ones from Artificial Analysis. sirloindenial (⬆12): Putting my lottery money on 1017. dryadofelysium (⬆17): This model is incredible. I had it build an Angular component showcase demo (my go-to test for new models), and the result was stunning (via DeepSeek Harness): https://x.com/D2KX_/status/2111461529304363165 shy_monkee (⬆27): Kimi has clearly been left in the dust, especially on benchmarks. By AI standards, that model is practically an ancient relic by now.