Arcee Trinity Mini: US-Trained Moe Model
1–10 of 16 posts
Re: Arcee Trinity Mini: US-Trained Moe Model
#2A moe model you say? How kawaii is it? uwu
Re: Arcee Trinity Mini: US-Trained Moe Model
#3A moe model you say? How kawaii is it? uwu
I hate that I laughed at this. Thanks ;)
Re: Arcee Trinity Mini: US-Trained Moe Model
#4Trinity Nano Preview: 6B parameter MoE (1B active, ~800M non-embedding), 56 layers, 128 experts with 8 active per token
Trinity Mini: 26B parameter MoE (3B active), fully post-trained reasoning model
They did pretraining on their own and are still training the large version on 2048 B300 GPUs
Re: Arcee Trinity Mini: US-Trained Moe Model
#5Looks like a less good version of qwen 30b3a which makes sense bc it is slightly smaller. If they can keep that effiency going into the large one it'll be sick.
Trinity Large [will be] a 420B parameter model with 13B active parameters. Just perfect for a large Ram pool @ q4.
Re: Arcee Trinity Mini: US-Trained Moe Model
#6A moe model you say? How kawaii is it? uwu
Capitalization makes a surprising amount of difference here...
Re: Arcee Trinity Mini: US-Trained Moe Model
#7A moe model you say? How kawaii is it? uwu
Meccha at present, but it may reach sugoi levels with fine-tuning.
Re: Arcee Trinity Mini: US-Trained Moe Model
#8> Trinity Large is currently training on 2048 B300 GPUs and will arrive in January 2026.
How long does the training take?
Re: Arcee Trinity Mini: US-Trained Moe Model
#9Excited to put this through its paces. It seems most directly comparable to GPT-OSS-20B. Comparing their numbers on the Together API: Trinity Mini is slightly less expensive ($0.045/$0.15 v $0.05/$0.20) and seems to have better latency and throughput numbers.
Re: Arcee Trinity Mini: US-Trained Moe Model
#10> Trinity Large is currently training on 2048 B300 GPUs and will arrive in January 2026. How long does the training take?
Couple days or weeks usually. No one is doing 9 month training runs