3 ms·
Luna is meant to compete with Haiku. What tasks are you seeing it equal Opus on?
by heaney-555 3mo ago
Luna is meant to compete with Haiku. What tasks are you seeing it equal Opus on?
- tosh 3mo agoluna is way better than haiku 4.5
- nateb2022 3mo agoI use Luna a lot (over 1T tokens since it came out) and I'd rank Luna (high/xhigh) on par with Sonnet 5, without hesitation.
- euazOn 3mo agoPer Artificial Analysis: - Haiku: 30 points - Luna Medium/High/Xhigh/Max: 38/46/49/51 points That's a massive difference: - 30 points is Gemma 4 31B territory - 50 points is GLM-5.2 (744B) territory.
- ignoramous 3mo agoCould OpenAI be running a different model for high+max?
- wahnfrieden 3mo agoNo OAI is just very good at reasoning models
- newtwilly 3mo agoAccording to the Artificial Analysis benchmark graph in the article, Luna can now outperform Sonnet 5 and Opus 5 low at ~4-10x less cost
- dannyw 3mo agoYou'd be surprised at what Luna can do, especially on xhigh or max. It's capable of working overnight, usually productively, just like Sol. Haiku 4.5, on the other hand, is comparable to performance to Gemma4 31B (with working tool call formatting) in my experience, and Gemma4 strongly wins on vision and multimodal.