Reddit asks: why run open-weight models in production instead of APIs?
allenling · reddit · 2026-11-09
A Reddit thread asks why teams choose open-weight models (author has been fine-tuning a Qwen vision model) over hosted APIs in production. Motivations cited: cost, privacy, self-hosting, avoiding vendor lock-in, and customization. The author wants input from people building real products or internal tools — as-is serving vs. fine-tuning, and whether self-managing models is worth the operational overhead.
More from Models
- Internal 32B/122B/256B checkpoints rumor debunked as AI-generated fake news — teortaxesTex · 2026-11-09
- Joseph Jacks: Liquid AI is the only US frontier lab keeping open weights ahead of China — JosephJacks_ · 2026-11-09
- Ant Group open-sources LemonDecider, a 4B decision model beating Jev 1.14 at 50.7ms median latency — aigclink · 2026-11-09
- Thinking Machines launches Inkling 2 Small: 594B-param MoE with 17B active, AA index 31 — FinanceYF5 · 2026-11-09
- Google reportedly building full Gemini 4 model suite, codenamed Krypton — ThePrimeNumbers · 2026-11-09
- Reddit users puzzled: why is everyone still on Qwen3 instead of the newer 3.5 models — opUserZero · 2026-11-09