Serving 4-bit Qwen3.8-27B on a single DGX Spark (GB10): 75 tok/s single-stream, 246 tok/s aggregate at 8-way concurrency. NVFP4 vs MixedInt4-AutoRound vs the FP8 baseline, measured on one harness — including why the quantization advantage collapses to +0.2% by c16.
☆51Sep 29, 2026Updated last month
Alternatives and similar repositories for Qwen3.8-27B-4-bit-on-a-single-DGX-Spark
Users that are interested in Qwen3.8-27B-4-bit-on-a-single-DGX-Spark are comparing it to the libraries listed below. We may earn a commission when you buy through links labeled 'Ad' on this page.
Sorting:
- A native vLLM 0.26 DGX Spark recipe for Qwen3.6 35B-A3B NVFP4, focused on high aggregate throughput, 262K context, and production-friendl…☆15Aug 31, 2026Updated 2 months ago
- Qwen3.8-27B on one DGX Spark GB10: DFlash 2 for 65 tok/s C1 and 120 C4; EAGLE 3/1/4 for ~191 C8, ~330 C16 and ~535 C32; native 262K, pinn…☆13Oct 6, 2026Updated last month
- ☆13Aug 27, 2026Updated 2 months ago
- Reproducible Qwen3.8-27B NVFP4 + DFlash2 recipe and qualification evidence for one NVIDIA DGX Spark☆30Aug 21, 2026Updated 2 months ago
- Run Qwen3.8-Flash-Next (180B) on a single DGX Spark by keeping its 51B n-gram embedding table on NVMe☆125Aug 30, 2026Updated 2 months ago
- Serverless GPU API endpoints on Runpod - Get Bonus Credits • AdSkip the infrastructure headaches. Auto-scaling, pay-as-you-go, no-ops approach lets you focus on innovating your application.
- Qwen3.8-27B NVFP4 (W4A16 shipped recipe) + MTP on NVIDIA DGX Spark GB10/SM121 — vLLM v0.27.2rc0, FP8 KV, 262K context, full reproducibili…☆27Aug 20, 2026Updated 2 months ago
- Custom inference recipes for NVIDIA DGX Spark — Qwen 122B DFlash hybrid☆92Updated this week
- Serve DeepSeek-V4-Flash-Vision EXL3 MixedK (full 256 experts) on one NVIDIA DGX Spark with vLLM — recipe, loader patches, and receipts☆14Sep 7, 2026Updated 2 months ago
- Model-agnostic LLM benchmarking + tuning in one command. A model is a 15-key profile file. Built in production on a 4x NVIDIA DGX Spark c…☆10Sep 16, 2026Updated last month
- Qwen3.8 27B on SGLang for DGX Spark☆481Sep 12, 2026Updated last month
- An environment where open models work as agents: a first-level harness with routing, skills, memory and sandboxed tools, on an engine tha…☆74Updated this week
- ☆15Sep 5, 2026Updated 2 months ago
- Qwen3.8-27B on one DGX Spark: SGLang + DFlash2 recipe, engine image, benchmarks, data and charts. By @redp314☆37Sep 6, 2026Updated 2 months ago
- DeepSeek v4 Flash EXL3 on one DGX Spark☆350Aug 24, 2026Updated 2 months ago
- Deploy to Railway using AI coding agents - Free Credits Offer • AdUse Claude Code, Codex, OpenCode, and more. Autonomous software development now has the infrastructure to match with Railway.
- vLLM inference solutions for NVIDIA GB10 (DGX Spark, sm121) — KIMI-K3 B12X_MLA + DSpark, GLM-5.2☆69Sep 30, 2026Updated last month
- Qwen3.8-27B NVFP4 on SGLang (DGX Spark / SM121): official image pin, native DSpark/EAGLE click-run, matched evals☆21Aug 19, 2026Updated 2 months ago
- First Int4-Int8Mix quantization of the full GLM-5.3 (743B), served on 4x NVIDIA DGX Spark (GB10) at TP4. Quantizer, verification gates, a…☆20Oct 6, 2026Updated last month
- Validated vLLM recipe: Qwen3.8-27B-NVFP4 at full 262,144-token context on a single RTX 5090, GPU-only (no CPU offload).☆16Aug 20, 2026Updated 2 months ago
- GLM-5.3-Flash EXL3 K2 2.33bpw on ONE DGX Spark: native MTP (layer 45, 288 experts), CUDA graphs on, full 262,144 ctx. Measured ~19 tok/s …☆10Sep 19, 2026Updated last month
- vLLM deployment for Unsloth Qwen3.6-35B-A3B-NVFP4-Fast on NVIDIA DGX Spark☆71Jul 29, 2026Updated 3 months ago
- Verified Qwen3.8 Flash-Next GGUF recipe for one NVIDIA DGX Spark or ASUS Ascent GX10☆60Oct 20, 2026Updated 2 weeks ago
- Serve Ornith-1.5-35B-A3B (NVFP4 + in-checkpoint MTP, 256k ctx) on one DGX Spark — vLLM b12x recipe plus CUDA-graph fixes.☆33Aug 19, 2026Updated 2 months ago
- AI reasoning skills distilled from 4,665 real Claude Fable 5 chain-of-thought traces. Mathematically tuned. Grade A (100%) emulation accu…☆13Jun 16, 2026Updated 4 months ago
- AI Agents on DigitalOcean Gradient AI Platform • AdBuild production-ready AI agents using customizable tools or access multiple LLMs through a single endpoint. Create custom knowledge bases or connect external data.
- vLLM recipe: GLM-5.3-Flash EXL3 K2 on one DGX Spark GB10. Native MTP k=2. Measured tok/s.☆26Sep 10, 2026Updated last month
- qwen38-flash-next recipe for solo GB10☆171Oct 30, 2026Updated last week
- Head-to-head comparison of local LLMs for agentic workflows (Hermes Agent, tool-eval-bench)☆36Oct 4, 2026Updated last month
- AEON vLLM Ultimate — vLLM 0.27.1 built from source for DGX Spark / Blackwell (sm_121a/GB10). DSpark quantized Markov heads, DFlash SWA on…☆149Oct 6, 2026Updated last month
- sparkDash ⚡ — Multi-DGX Spark Monitoring Dashboard☆839Updated this week
- Qwen3.6-27B (NVFP4) DFlash for DGX Spark☆22Jul 31, 2026Updated 3 months ago
- Qwen 3.8 Flash NVFP4 with PLE NVFP4 quantized to run on a single DGX Spark☆15Aug 28, 2026Updated 2 months ago
- Qwen3.8-Flash-Next (NVFP4) on DGX Spark in vLLM: one Spark 43.9 tok/s with our disk-backed n-gram table patch, staged gather and reduced-…☆137Sep 7, 2026Updated 2 months ago
- ☆16Mar 8, 2026Updated 7 months ago
- 1-Click AI Models by DigitalOcean Gradient • AdDeploy popular AI models on DigitalOcean Gradient GPU virtual machines with just a single click. Zero configuration with optimized deployments.
- Qwen3.8 Flash Next on one DGX Spark (TensorFold)☆383Oct 30, 2026Updated last week
- Qwen3.8-27B ADay777-Ablit · NVFP4 · A4Q · NVFP4-KV (4M+ pool) · MTP-3 — single DGX Spark (GB10/sm121), eugr vLLM 0.27. 1M ctx @ c=4.☆24Aug 17, 2026Updated 2 months ago
- Serve LibertAIDAI/GLM-5.3-Flash-NVFP4 on 2x NVIDIA DGX Spark (GB10) with patched vLLM TP=2, 262k context, MTP-4.☆21Updated this week
- Full GLM-5.3 (753B, Int4-Int8Mix) on 4x NVIDIA DGX Spark, vLLM TP4☆23Oct 13, 2026Updated 3 weeks ago
- DeepSeek V4 Flash DSpark 1M NVFP4 KV recipe for 2x DGX Spark☆496Sep 10, 2026Updated last month
- ☆54Sep 13, 2026Updated last month
- 1 spark glm 5.3 flash☆11Sep 7, 2026Updated 2 months ago