WanGP brings explorable world models to 8GB GPUs, near-instant generation on RTX 3080
cocktailpeanut · x · 2026-10-27
WanGP v18.81 brings world model generation to GPUs with as little as 8GB VRAM:
- Build a world from text (Hy-Nomas) or from an image (EVOKE), then explore it with your keyboard; save progress, record journeys, and revisit places
- A tester ran the low-res EVOKE model on an RTX 3080 (10GB): results appeared within seconds of downloading, with an 8-9 second generation pause on every move/look change — rough quality but in a fun, up-and-coming way
- The bundled local agent "Deepy Prime" gets roughly 2x context at the same VRAM, plus new abilities: web search for compatible LoRAs, Python-driven graphics/animation/batch workflows, 3D asset and explorable world generation, and video restore/upscale/outpaint via Media Flow
Free to run locally via Wan2GP/Pinokio.
Related event: WanGP v18.81 Brings Tencent's Hy-Nomas World Model to 8GB GPUs(4 posts)→
More from Infra
- Sail equivalence proof and riscv-coq cross-checks uncover real RISC-V spec bugs — xennygrimmato_ · 2026-10-27
- xAI's Colossus 1 hits ~14.6 exaflops FP64, ~5x Top500 leader — cis_female · 2026-10-27
- Lambda runs 19 B200 nodes in a 16-node power budget, boosting throughput 24% — TheZachMueller · 2026-10-27
- Podcast: Micron's HBM roadmap, Meta's AI infra, and NVIDIA's supply chain lock-in — BenBajarin · 2026-10-27
- SGLang MegaKernel + PD Pushes the TCO-Latency Frontier on GB300, SemiAnalysis Benchmarks Show — AccBalanced · 2026-10-27
- Scaling self-hosted Convex: compiling TypeScript Convex functions to WASM — leventov · 2026-10-27