---
type: "Article"
title: "Ornith-1.5-122B EXL3 2,58 bpw mit vLLM auf 1× DGX Spark"
description: "Ornith-1.5-122B EXL3 2,58 bpw mit vLLM auf 1× DGX Spark: Rezept von vcruz305 mit Tempo, Bedingung und Quelle je Zahl, Gewichten und Lizenz."
resource: "https://www.contextstudios.ai/de/lokale-ki/vcruz305--ornith-15-122b-exl3-2.58-1spark-vllm"
language: "de"
generated:
  by: "process:contextstudios-md/1"
  at: "2026-11-11T03:38:03.171Z"
status: "stable"
---

# Ornith-1.5-122B EXL3 2,58 bpw mit vLLM auf 1× DGX Spark

Ornith-1.5-122B EXL3 2,58 bpw mit vLLM auf 1× DGX Spark: 30,5 tok/s laut github.com (Datensatz-Stand: 3. Nov. 2026).

- Kreator: [vcruz305](https://www.contextstudios.ai/de/lokale-ki/kreatoren/vcruz305.md)
- Engine: vLLM 0.32.0 + vllm-exl3 Fork 3275819 (-DP2B_CB=2), VLLM_EXL3_COOP=1 + COALESCE_SHARDS=1, BF16-SSM
- Quantisierung: EXL3 SAGE 2.58bpw (MixedK, mul1-Codebook, lm_head K=8), BF16 KV
- Modellfamilie: Ornith-1.5-122B
- Hardware: dgx-spark x1
- Kontext: 262144

## Alle Zahlen

- 30,5 tok/s (Spitze; Bestwert der Quelle, 1 Anfrage, Prompt-Typ nicht genannt, greedy, 512 Token, Standard-serve.sh ohne MTP) — [source](https://github.com/vcruz305/Ornith-1.5-122B-EXL3-DGX-Spark-recipe#measured-performance-on-1-nvidia-dgx-spark-gb10-sm_121-128-gb-unified)
- decode_tps: 30.51 (1 Anfrage) — [source](https://github.com/vcruz305/Ornith-1.5-122B-EXL3-DGX-Spark-recipe#measured-performance-on-1-nvidia-dgx-spark-gb10-sm_121-128-gb-unified)
- prefill_tps: 1119.1 (1 Anfrage) — [source](https://github.com/vcruz305/Ornith-1.5-122B-EXL3-DGX-Spark-recipe#measured-performance-on-1-nvidia-dgx-spark-gb10-sm_121-128-gb-unified)
- ttft_ms: 1830.1 (1 Anfrage) — [source](https://github.com/vcruz305/Ornith-1.5-122B-EXL3-DGX-Spark-recipe#measured-performance-on-1-nvidia-dgx-spark-gb10-sm_121-128-gb-unified)

## Gewichte

- [vcruz305/Ornith-1.5-122B-EXL3-SAGE-2.58bpw](https://huggingface.co/vcruz305/Ornith-1.5-122B-EXL3-SAGE-2.58bpw) (main, mit)
- [ornith-ai/Ornith-1.5-122B](https://huggingface.co/ornith-ai/Ornith-1.5-122B) (base, mit)

[https://github.com/vcruz305/Ornith-1.5-122B-EXL3-DGX-Spark-recipe](https://github.com/vcruz305/Ornith-1.5-122B-EXL3-DGX-Spark-recipe)

## Related

- [Lokale KI auf Ihrer Hardware.](https://www.contextstudios.ai/de/lokale-ki.md)