SenseNova-Loop-9B Hits 1,300+ tok/s on a Laptop RTX 5090, Benchmarked vs Qwen 27B
tebjan · reddit · 2026-11-22
A Reddit user benchmarks SenseNova-Loop-9B (looped transformer, NVFP4 + GPTQ error compensation) at over 1,300 aggregated tok/s on a laptop RTX 5090. vs Qwen3.8-27B at one-third the size: on par on GSM8K (95.4) and multi-turn tool calls; 4-5 points behind on HumanEval and single tool calls; clearly behind on hard multi-step tool scenarios (-10.5), needle-in-haystack (-16.7) and MMLU-Pro (-21.3). Good for fast routine agent work, not hard tasks.
More from Infra
- Gemini 4 Neon autonomously writes GPU kernels, 5.9x faster prefill on AMD hardware — sebkrier · 2026-11-22
- Intel's 350W D270 claims 5.9x lead over RTX PRO 6000 — but the fine print reveals benchmark games — StefanoGogioso · 2026-11-22
- A tale of two Nvidias: solid earnings vs. execs calling for Jensen's return — TechCrunch AI · 2026-11-22
- Chollet: GPU spot prices keep rising as HBM sells out through 2029 — AI compute buildout isn't over — fchollet · 2026-11-22
- DeepMind bakes speculative decoding into Gemma 4: up to 3.1x server speedups, 2.2x on smartphone GPUs — dair_ai · 2026-11-22
- Gavin Baker: No investor feels misled by Nvidia's China GPU disclosures — GavinSBaker · 2026-11-22