FR
Réserver un appel

Local AI · Recipe · 1× DGX Spark

GLM-5.3-Flash EXL3 2.25 bpw with ExLlamaV3 on 1× DGX Spark (256k context)

GLM-5.3-Flash EXL3 2.25 bpw with ExLlamaV3 on 1× DGX Spark (256k context): 20.6 tok/s according to github.com (dataset as of Sep 29, 2026).

by r0b0tlab

20.6tok/sEveryday

1 request, prose prompt, with DFlash2 K=5 (per recipe)

Source: github.com
Intelligence (original model) · Artificial Analysis42with thinking

≤ 3 bit: quantization may cost quality

Instructions in the README Repo

Engine

Engine
ExLlamaV3 v1.5.2 + TabbyAPI 816c321 (ghcr.io/r0b0tlab/glm53-flash-exl3-exllamav3-gb10:b25c467)
Quantization
EXL3 2.25 bpw (Drafter EXL3 3.00 bpw), FP16 KV
Model family
GLM-5.3-Flash
Context
262,144
Creator
r0b0tlab
GitHub stars
4
Repo updated
Sep 28, 2026

Measurements

Every sourced value of this recipe, each with its condition and source. Bars relative to the largest value in the group.

Decode · 1 request

  • Everyday (decode)20.6 tok/scontext 8K-Profil · Matched 8K C1, Median aus 5; Prosa regrediert unter K=5 (AR 22.54) Source 
  • Peak (decode)52.9 tok/scontext 8K-Profil · Structured, DFlash2 K=5, matched 8K C1, Median aus 5; Code 49.96 Source 

What you need

Hardware
1 × NVIDIA DGX Spark (GB10)
Weights
Engine
ExLlamaV3 v1.5.2 + TabbyAPI 816c321 (ghcr.io/r0b0tlab/glm53-flash-exl3-exllamav3-gb10:b25c467)
Context
262,144 tokens

Notes

What matters before you rebuild it.

  • ≤ 3 bit: quantization may cost qualityAt 3 bit and below the model may answer noticeably worse than the original. The intelligence number refers to the original.
  • Custom kernel required
  • vision
  • tool-calling
  • Non-commercial

Sources

Related recipes

2× DGX Spark

GLM-5.3-Flash EXL3 with TensorFold on 2× DGX Spark (1M context, 4 streams)

by MiaAI-Lab

60.4tok/sEveryday

Prose decode, 1 request; quality (FP8 KV, 4-bit dense): GSM8K 98.8% / HumanEval 95.7% / HumanEval+ & MBPP+ 86.5% Source 

Intelligence42with thinking
  • ≤ 3 bit: quantization may cost quality
  • Custom kernel required
  • Non-commercial
  • License unclear

EXL3/TR3 4bpw routed experts + 4-bit dense, BF16 elsewhere, FP8 KVTensorFold v0.6.0 + Recipe v1.10 (README: 96 Patches; gepinntes Image v0.6.0-e66c657a433d mit 107 Patches: DFlash2+Copy-Drafts, 4-bit dense, FP8-KV, RoCE-One-Shot-All-Gather, Shared KV-Pool, optionaler Spill-Tier, Vision, Tool-Calling, Anthropic /v1/messages; PARALLEL=8 opt-in)Weights Repo Updated Oct 10, 2026

Details
2× DGX Spark

GLM-5.3-Flash 320B-A18B EXL3 4 bpw with ExLlamaV3 on 2× DGX Spark

by MiaAI-Lab

36.1tok/sEveryday

1 request, prose prompt, with DFlash2 (per recipe) Source 

Intelligence42with thinking
  • ≤ 3 bit: quantization may cost quality
  • Custom kernel required
  • Non-commercial
  • License unclear

EXL3 4bpw (TR3) + FP8 KV (fp8_ds_mla)TensorFold-basiertes Kit (ghcr.io/miaai-lab/glm-5.3-flash-2x-dgx-sparks:exl3-instanttensor), DFlash2 7 draft tokens; opt-in Speed-Boost-Profil (einmaliger ~25-min-Build auf dem Head)Weights Repo Updated Oct 4, 2026

Details

← Back to overview

Local AI in your company?

In a workshop we work out which models and which hardware fit your tasks, and build the first agent on your infrastructure.

See the AI workshopBack to overview