tensorfold

keys-Wasserman Slate AI Studio powered by TensorFold: an AI film studio for Mac and NVIDIA DGX Spark. Plan a film like a crew (Slate workflow by Sam Wasserman), render Qwen-Image-2.1 images Turbo and MiniMax H3 / FastH3 video with sound on TensorFold.

NVFP4 BIZ: nvidia/GLM-5.3-Flash-NVFP4 as distributed, on DGX Spark-class GB10 systems. 2.x serves it with TensorFold (TP=2 or TP=3, FP8 KV, MTP, drafted replies equal serial); 1.x with a pinned vLLM (TP=2 or TP=3, images, optional AXL repack). Apache-2.0 code, MIT weights fetched separately. BIZ = business-use intent, not support or certification.

GLM-5.3-Flash EXL3 on 4x RTX PRO 6000 with TensorFold: one-command recipe and engine patches. By Aevonix Research in collaboration with Mia's AI Lab.

MiniMax H3 audio-video model on TensorFold NVFP4 kernels: a faster ComfyUI loader for one RTX 50-series GPU (2.3x stock at 20 steps, 81 s per 5 s 1344x768 video with Turbo + sparse attention on an RTX 5070 Ti)

Qwen3.8-Flash-Next EXL3 3.05bpw on ONE DGX Spark (GB10): ~2.4x prefill (700->1777 t/s), prompt-lookup drafter, native vision/video, 262k ctx. Unofficial TensorFold v0.6.5 fork, 18 patches, byte-exact receipts.

GLM-5.3-Flash EXL3 on 2x RTX PRO 6000 with TensorFold: one-command recipe and engine patches. By Aevonix Research in collaboration with Mia's AI Lab.

GLM-5.3 EXL3 on 4x RTX PRO 6000 with TensorFold: one-command recipe and engine patches. By Aevonix Research in collaboration with Mia's AI Lab.

DeepSeek-V4.1-Flash-1017 on two NVIDIA DGX Sparks (GB10) with TensorFold TP2: scripts, chat-template fix, speed and quality results

Kolibri-1 (Aleph Alpha) on two NVIDIA DGX Sparks with TensorFold: ~2x decode tok/s vs one Spark. Recipe, patches, measured results.

ThinkingCap EXL3 with pinned TensorFold CUDA: measured DGX Spark profile and guarded experimental 24 GB NVIDIA GPU qualification path.

TensorFold deployment for GLM-5.3-Flash on 4x NVIDIA CMP 170HX (GA100 sm_80): bit-exact 4-rank TP, shm all-gather, W4A16 affine - complete tested solution (199.4 tok/s structured, exactness maintained)

Keep every AI model on a NAS; insert any of them on your GPU box with one click (Ollama / vLLM) — never evicts a model you didn't name. Stdlib Python, demo mode, web UI.

Four DGX Spark switchless TensorFold TP4 deployment for GLM-5.3-Flash EXL3, with ring exchange and measured benchmarks

Qwen3.8 Flash Next on one NVIDIA Jetson AGX Thor with TensorFold: recipe, gated benchmarks, and a head-to-head with a DGX Spark

DeepSeek-V4.1-Flash-1017 on native TensorFold, 2x DGX Spark (TP=2): prose 55 / code 104 tok/s at 1 stream, 135 / 203 at 8, 0.2 s first token · text, images, video · EXL3 2.9 bpw

TensorFold v0.6.0 GLM-5.3-Flash engine generalized from 2 to 4 DGX Sparks (TP4) over a switchless RoCE fiber ring - uncensored EXL3 checkpoint, overlay + tooling. Upstreams: ashhart/TensorFold, MiaAI-Lab 2x-Spark stack, alexellis/switchless-nccl.

GLM-5.3-Flash on four switchless DGX Sparks with Mia TensorFold, prepared RoCEnante and patched NCCL: builds, recovery, qualification and benchmarks.

Qwen3.8-Flash-Next INT4-Mixed on native TensorFold, 2x DGX Spark (TP=2): prose 78 / code 127 tok/s at 1 stream, 304 / 516 at 16, 0.04 s first token · text, images, video, tools, 1M context · AutoRound INT4

Reproducible MLX Q4 conversion of Cantina Security Apex Flash 1 for TensorFold on dual DGX Spark

Self-built EXL3 4 bpw TensorFold pack of RED-SNOW-5.3-Flash for 2x DGX Spark: build recipe, serving kit, measured results