Gradio distills FLUX.3 Klein's 9B prompt upsampler into 0.8B and 2B models that run on CPU
Gradio · x · 2026-10-22
Gradio built a Prompt Upsampler Arena with gr.Workflow to compare four approaches side by side: raw short prompts, the official 9B prompt upsampler, and two tiny distilled planners (0.8B and 2B).
They distilled FLUX.3 Image Klein's 9B prompt upsampler into 0.8B and 2B models: a short prompt goes in, a full prompt plus bounding boxes come out. The 0.8B version can even run on a CPU. Sample outputs show the distilled planners rival the 9B default on layout and quality with faster planning times.
Related event: Gradio distills FLUX.3 prompt upsampler to 0.8B, runs on CPU(3 posts)→
More from Infra
- After MoE and MTP, local dense models haven't seen architectural speedups — Express_Quail_1493 · 2026-10-22
- NVIDIA intros DSX PODs at GTC Berlin, up and running in 14 days — chemist_slime · 2026-10-22
- Deutsche Bank Deploys Farsight On-Prem as Financial AI Question Shifts to Where Models Run — huggingface · 2026-10-22
- Securing coding agents with Docker Sandboxes and edge LLM inference at ATO — HowDevelop · 2026-10-22
- Revolut's massive outage blamed on GCP, but ex-sales contact says it rejected multi-zone redundancy — thedealdirector · 2026-10-22
- 0.9B Model Transcribes a 4-Person Meeting on a MacBook With 0.08% Speaker Confusion — Binary_orchid · 2026-10-22