Local AI · Recipe · 4× RTX PRO 6000

GLM-5.3-Flash TrellisMX codec-v2 FP8 with vLLM on 4× RTX PRO 6000

GLM-5.3-Flash TrellisMX codec-v2 FP8 with vLLM on 4× RTX PRO 6000: 189 tok/s according to huggingface.co (dataset as of Nov 3, 2026).

by brandonmusic

189tok/sPeak

1 request, context 0 (78 prompt tokens), synthetic sustained decode with ignore-EOS, with MTP3 (per recipe)

Source: huggingface.co

Intelligence (original model) · Artificial Analysis42with thinking

Install: GLM-5.3-Flash TrellisMX codec-v2 FP8 with vLLM on 4× RTX PRO 6000

Command

hf download brandonmusic/glm-5.3-flash-tr3-fp8 --revision codecv2-prefill-20260927 --local-dir /absolute/path/to/glm-5.3-flash-tr3-fp8

Full instructions in the README Repo

Engine: vLLM (TrellisMX-Runtime, Image verdictai/trellismx:glm53-codecv2-prefill-20260927 @sha256:6240b05f…; TP4/DCP4, MTP3)
Quantization: TrellisMX codec-v2 K4 (4,25 bpw Payload, Routed- + MTP-Experten) → native FP8 E4M3 Experten-Math (UE8M0/K32), FP8 KV
Model family: GLM-5.3-Flash
Context: 1,000,000
Creator: brandonmusic
GitHub stars: 4
Repo updated: Oct 29, 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

Peak (decode)189 tok/scontext 0 · Owner-Benchmark run-inference-bench v0.4.29: 30-s-Messfenster, 3 s Warmup, max. 8.192 Output-Token, ignore-EOS, Default-Sampling (Temperatur 1,0 / top-p 0,95), MTP3, FP8 KV; eine Serie ohne Wiederholungen Source

Prefill by context

32K9,520 tok/sStandalone-Prefill, Client-Messung (Prompt-Token / Client-TTFT), 32.321 tatsächliche Prompt-Token, eindeutige Präfixe gegen Cache-Reuse, Median aus 3 Samples Source

Time to first token

Shorter is better.

Prompt not stated13,751 mscontext 128K · Client-TTFT, 128.880 tatsächliche Prompt-Token, ein Sample Source

What you need

Hardware: 4 × NVIDIA RTX PRO 6000 Blackwell
Weights: Main weightsbrandonmusic/glm-5.3-flash-tr3-fp8 MIT
Basezai-org/GLM-5.3-Flash-BF16 MIT
Engine: vLLM (TrellisMX-Runtime, Image verdictai/trellismx:glm53-codecv2-prefill-20260927 @sha256:6240b05f…; TP4/DCP4, MTP3)
Context: 1,000,000 tokens

Notes

What matters before you rebuild it.

Custom kernel required

Sources

Related recipes

4× DGX Spark

GLM-5.3-Flash NVFP4/MXFP8 CSF with SparkRing on 4× DGX Spark

by fujitsupolycom

68tok/sEveryday

1 request, generation at 32K context (speed table; measured only as a path of four Sparks inside a ring of eight, not as a ring of four), thinking always on (measured before the profile default high, i.e. template default max), with MTP3 (per recipe) Source

Intelligence42with thinking

Custom kernel required
License unclear

NVFP4/MXFP8 CSF QAD (Local Inference Lab)vLLM (local-inference-lab Fork) via SparkRing (SIRCL-Collectives, switchless RDMA ueber ConnectX-7), Image Release 2026.10.2Weights Repo Updated Nov 2, 2026

Details

4× DGX Spark

GLM-5.3-Flash EXL3 4bpw with TensorFold on 4× DGX Spark

by tonyd2wild

65.4tok/sEveryday

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

Intelligence42with thinking

Custom kernel required
Non-commercial
License unclear

EXL3 4bpw routed experts (mcg) + q4 dense (at load) + FP8 KVTensorFold v0.6.0 + MiaAI-Lab GLM-5.3-Flash patch series (engine-multi, + tonyd2wild changes)Weights Repo Updated Oct 12, 2026

Details

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 25, 2026

Details

2× DGX Spark

GLM-5.3-Flash EXL3 4bpw with TensorFold on 2× DGX Spark

by tonyd2wild

44tok/sEveryday

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

Intelligence42with thinking

Custom kernel required
Non-commercial
License unclear

EXL3 4bpw routed experts (mcg) + q4 dense (at load) + FP8 KVTensorFold v0.6.0 + MiaAI-Lab GLM-5.3-Flash patch series (engine-multi, + tonyd2wild changes)Weights Repo Updated Oct 12, 2026

Details

4× DGX Spark

GLM-5.3-Flash 320B-A18B NVFP4 with vLLM on 4× DGX Spark

by tonyd2wild

40.8tok/sEveryday

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

Intelligence42with thinking

Custom kernel required
Non-commercial

NVFP4 weights + FP8 KV (Lane B: NVFP4 KV)vLLM 0.28.0Weights Repo Updated Oct 10, 2026

Details

3× DGX Spark

GLM-5.3-Flash 320B EXL3 4 bpw with vLLM on 3× DGX Spark

by MiaAI-Lab

40.1tok/sEveryday

1 request, prose prompt, with DFlash2, TR3 weights (per recipe) Source

Intelligence42with thinking

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

EXL3/TR3 4 bpw (routed experts), FP8 KV (fp8_ds_mla)vLLMWeights Repo Updated Oct 18, 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 workshop Back to overview

Engine

Measurements

Decode · 1 request

Prefill by context

Time to first token

What you need

Notes

Sources

Related recipes

GLM-5.3-Flash NVFP4/MXFP8 CSF with SparkRing on 4× DGX Spark

GLM-5.3-Flash EXL3 4bpw with TensorFold on 4× DGX Spark

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

GLM-5.3-Flash EXL3 4bpw with TensorFold on 2× DGX Spark

GLM-5.3-Flash 320B-A18B NVFP4 with vLLM on 4× DGX Spark

GLM-5.3-Flash 320B EXL3 4 bpw with vLLM on 3× DGX Spark

Local AI in your company?