Macaron V1 reportedly ships in 748B and 50B variants, with GLM-5.2 and Qwen 3.6 post-training
AdinaYakup · x · 2026-07-27
A reply shares details about Macaron V1, which reportedly comes in two variants:
- Venti: a 748B flagship model, built from 744B base + 4×1B LoRA specialists, post-trained on GLM-5.2.
- Tall: a 50B local-deployment version, built from 35B base + 4×3.7B LoRA specialists, post-trained on Qwen 3.6.
The post says the model and harness were co-designed with MinT and MindForge, all built in-house by Mind Lab.
Related event: Macaron V1 Released with 748B and 50B Variants(2 posts)→
More from Models
- theo builds his own visualizer for today's agent models, showing how cheap Luna really is — ivan_bezdomny · 2026-09-23
- Why ChatGPT Still Wins: One User's Split Between Muse, Claude and Codex — mobileraj · 2026-09-23
- Muse reportedly offers 4B tokens/week for ~$100/month, sparking industry price-disruption talk — NewYak4281 · 2026-09-23
- GPT-6 Sol and Luna appear in OpenAI docs, alongside guidance on reasoning effort — cedric_chee · 2026-09-23
- GPT-6 tested on LIBERO robot task: turns on stove, fails to grasp moka pot — YuXiang_IRVL · 2026-09-23
- Ternary Bonsai 2 27B: 5.9GB weights retain ~95% of full-precision reasoning — cephaloform · 2026-09-23