300M-param image model trained for €5000 claims SD 1.5-level performance
incorporo · reddit · 2026-09-29
Logolabs released Agate-003-preview, a <300M parameter (including text encoder & VAE) image generation model claiming SD 1.5-level performance at 3x fewer parameters, trained for under €5000 in 630 GH-200 GPU-hours (319 GPU-h for the core run).
Key techniques:
- Planner + renderer architecture: a transformer planner works at low resolution with cross-attention to text tokens, feeding a fully convolutional renderer with strong image priors
- Weighted loss using P-Dino/LPIPS and saliency, prioritizing faces and text over backgrounds
Lineage: Agate-001 → 002 (256×256) → 003 (native 512×512 with learned down/up-samplers keeping planning resolution fixed). The tiny model supports cheap, accessible LoRA training; the team plans to use it for font/glyph/symbol generation. Available on Hugging Face, with a WebGPU demo and ComfyUI nodes.
Related event: Logolabs Unveils Agate, a 260M-Parameter Text-to-Image Model Rivaling SD1.5(2 posts)→
More from Multimodal
- Testing Qwen Image 2.1 and Minimax H3 on anime-to-photo conversion and film photo restoration — Ant_6431 · 2026-11-16
- Second take of a 'Cubee run' video generated with Minimax H3 — JahJedi · 2026-11-16
- Non-artist builds 14-chapter hand-painted storybook game with Fable 5.2 in 4 hours — jradoff · 2026-11-16
- Subject LoRAs on Minimax H3 Image learn poor likenesses while identical datasets excel on Qwen 2.1 — Ad_Organic · 2026-11-16
- Seedance-generated anime train scene stuns Reddit with cinematic quality — Ok-Solution-7889 · 2026-11-16
- AI-generated 'Men in Black' reimagined in 1977 goes viral — techhalla · 2026-11-16