DeepReinforce

Ornith 1.0 35B A3B

Name: Ornith 1.0 35B A3B
Rating: 81 (87 reviews)
Author: DeepReinforce

Frontier

HuggingFace

231.3KDownloads348LikesJun 2026Veröffentlicht262K TokenKontextMITLizenz72 StarkQualität

Ornith 1.0 35B A3B (35.099998474121094B parameters) requires approximately 23.2 GB of VRAM with Q4_K_M quantization. As a Mixture of Experts model with 3B active parameters, it uses less memory than its total parameter count suggests. For the best balance of quality and speed, we recommend hardware with at least 27 GB of VRAM.

Loslegen

— kopieren & einfügen, um lokal auszuführen

Copy-paste commands to run Ornith 1.0 35B A3B on your machine.

Run

docker run --rm -it ghcr.io/ggerganov/llama.cpp:full \
  --hf-repo "deepreinforce-ai/Ornith-1.0-35B" \
  --hf-file "Ornith-1.0-35B-Q4_K_M.gguf" \
  -c 4096 -ngl 99

Quick specs

Parameters35.1B (3B active)

Architecturemoe (MoE)

Context262K tokens

Modalitytext

Min RAM13.7 GB

Rec. RAM21.4 GB (Q4_K_M)

LicenseMIT

FamilyOrnith

✓ Code✓ Reasoning

About this model

Ornith-1.0-35B is DeepReinforce's lightweight self-improving coding agent, built on Qwen 3.5. A 35B-total MoE (256 experts, 8 active, ~3B activated per token) with a hybrid linear + full-attention backbone, designed for efficient single-GPU agentic deployment with a 262K context.

•Self-improving agentic RL: jointly refines solution generation and reasoning scaffolds.
•35B MoE with only ~3B activated per token — the lightweight member of the Ornith family.
•Hybrid linear-attention + full-attention (1-in-4 layers) for reduced KV-cache cost.
•262K context, tuned for single-GPU agentic coding.

Schnellauswahl

Bestes BudgetA

Mac mini M4 64GB~$1,099 — 13 tok/s

Beste GesamtwahlS

RTX 5090 32GB~$1,999 — 139 tok/s

Beste Hardware

Top-Empfehlungen für Ornith 1.0 35B A3B

RTX 5090 32GBS

32 GB

RTX PRO 4500 Blackwell 32GBS

32 GB

AMD Instinct MI100 32GBS

Dieses Modell ausführen

Ornith 1.0 35B A3B on RTX 5090 32GB Ornith 1.0 35B A3B on RTX PRO 4500 Blackwell 32GB Ornith 1.0 35B A3B on AMD Instinct MI100 32GB

Quantisierungsoptionen

VRAM-Schätzungen nach Quantisierungsstufe

No hardware detected — fit column shows raw VRAM estimates

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	13.7 GB	Low	—
Q3_K_S	3	17.2 GB	Low	—
NVFP4	4	19.7 GB	Medium	—
Q4_K_M	4	21.4 GB	Medium	—
Q5_K_M	5	25.3 GB	High	—
Q6_K	6	28.8 GB	High	—
Q8_0	8	37.6 GB	Very High	—
F16	16	72.0 GB	Maximum	—

Hardware-Kompatibilität

Eignungsschätzungen für alle Hardware

Rechner öffnen

Computing compatibility...

Speicheraufschlüsselung

Reference: RTX 2060 6GB

Weights21.4 GB

KV Cache0.3 GB

Runtime0.9 GB

Headroom0.6 GB

Häufig gestellte Fragen

FAQ — Ornith 1.0 35B A3B

Siehe auch

Quantisierungsleitfaden Bewertungsmethodik Rechner öffnen