llmrun
orcarouterOrcaQwen3_5ForCausalLM

OrcaSAQ 2 27B — Hardware Requirements & GPU Compatibility

ChatReasoningFunctions

OrcaSAQ 2 27B is a 6.8B-parameter open language model from orcarouter in the Orca family. It supports a context window of up to 262,144 tokens. At BF16 it needs about 14.29 GB of VRAM — see which GPUs and Macs can run it below.

6.8B params · dense262K contextApache 2.05.1K downloads279 likes
Based on Qwen3.8 27B
A

Runs well on a 24 GB GPU

Runs from16 GB BF16 · 32KBest on24 GB BF16 · ~46 tok/s · 128KMacs from18 GB BF16 · ~7 tok/s

Apache 2.0

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run OrcaSAQ 2 27B

One recommended way. Every other method, OS and quant is in the step-by-step guide.Each way opens a step-by-step guide for your OS and quant.

Recommended

vLLM
one command · needs vLLM installed
$vllm serve orcarouter/OrcaSAQ-2-27B --max-model-len 16384

vLLM repo with its own quantisation

Download: Weights on Hugging Face

install, OS, other quants

Ways to runOther ways

each opens step-by-step

Run in cloud

Runs on RTX 3090 (24 GB) (9 GB headroom) · BF16

Generation speed
~43 tok/s
generation speed
Cost per 1M output tokens
$1.28
per 1M output tokens

How Much VRAM Does OrcaSAQ 2 27B Need?

Select a quantization level; the listed hardware adheres to it.

QuantizationVRAM+32KOn RTX 4090 (reference)
14.3 GB15.6 GB16-bit · bfloat — Brain floating point 16 — preferred for trainingFits: 128K context · ~46 tok/s
est.calculated, no published file yet

Selected BF16

Brain floating point 16

14.7GBat 16K context

Weights
13.5 GB
KV cache
0.9 GB
Overhead
0.3 GB

Fits RTX 4090 (reference) · 9.3 GB free · ~46 tok/s

≥50 tok/s needs ≥1.10 TB/s · RTX 5090 ~82, M5 Ultra ~60

Which Hardware Can Run OrcaSAQ 2 27B?

BF16 · 14.3 GB

OrcaSAQ 2 27B (BF16) requires 14.3 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 19+ GB is recommended. Using the full 262K context window can add up to 13.9 GB, bringing total usage to 28.2 GB. 30 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti, NVIDIA GeForce RTX 5080. 139 devices with unified memory can run OrcaSAQ 2 27B, including Mac Studio M5 Ultra (256 GB), Mac Studio M5 Ultra (512 GB), Mac Mini M5 Pro (24 GB).

By VRAM tier · typical GPUs and Macs at BF16
8–12 GB · RTX 4060, RTX 3060 12GBDoesn't fit — smallest quant needs 14.7 GB8 GB tier →12 GB tier →
16 GB · RTX 5060 Ti 16GBBF16 · 14.7 GB · ~20 tok/s · 32K16 GB tier →
24 GB · RTX 4090referenceBF16 · 14.7 GB · ~46 tok/s · 128K24 GB tier →
Mac 32 GB · Mac Mini M4BF16 · 14.7 GB · ~6 tok/s · 256KMacs →
Mac 64 GB · Mac Studio M4 MaxBF16 · 14.7 GB · ~27 tok/s · 256KMacs →

Alternatives to OrcaSAQ 2 27B

Orca family · 9 models →

Similar models

6
ModelScoreVRAMtok/sWhy it is similar
Olmo 3 7B InstructAllen AI—8.4 GB~114same size
Mathstral 7B v0.1Mistral AI—6.8 GB~133same size
Mistral 7B Instruct v0.3Mistral AI—6.8 GB~133same size
CodeQwen1.5 7BAlibaba—5.7 GB~137same size
Starling LM 7B BetaNexusflow—5.7 GB~133same size
Llava 1.5 7B HFllava-hf—5 GB~141same size

Specifications

Publisher
orcarouter
Family
Orca
Parameters
6.8B
Architecture
Qwen3_5ForCausalLM
Context Length
262,144 tokens
Vocabulary Size
248,320
Release Date
2026-09-24
License
Apache 2.0

Frequently Asked Questions

How much VRAM does OrcaSAQ 2 27B need?

OrcaSAQ 2 27B requires 14.3 GB of VRAM at BF16. Full 262K context adds up to 13.9 GB (28.2 GB total).

VRAM = Weights + KV Cache + Overhead

Weights = 6.8B × 16 bits ÷ 8 = 13.5 GB

KV Cache + Overhead ≈ 0.8 GB (at 2K context + ~0.3 GB framework)

Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.

KV Cache + Overhead ≈ 14.7 GB (at full 262K context)

VRAM usage by quantization

14.3 GB
28.2 GB

Learn more about VRAM estimation →

Can I run OrcaSAQ 2 27B on a Mac?

Yes — MacBook Pro 14" M3 Pro (18 GB) and 117 other Macs can run OrcaSAQ 2 27B. Apple Silicon uses unified memory, so the model shares RAM with the system. At BF16 you need at least 14.3 GB of usable unified memory (RAM minus macOS overhead).

Can I run OrcaSAQ 2 27B locally?

Yes — OrcaSAQ 2 27B can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is OrcaSAQ 2 27B?

At BF16, OrcaSAQ 2 27B can reach ~336 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~46 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.

tok/s = (bandwidth GB/s ÷ model GB) × efficiency

Example: NVIDIA B200 → 8000 ÷ 14.3 × 0.65 = ~364 tok/s

Estimated speed at BF16 (14.3 GB)

~364 tok/s
~46 tok/s
~364 tok/s
~336 tok/s

Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.

Learn more about tok/s estimation →

What's the download size of OrcaSAQ 2 27B?

At BF16, the download is about 13.54 GB.

Which GPUs can run OrcaSAQ 2 27B?

30 consumer GPUs can run OrcaSAQ 2 27B at BF16. Top options include AMD Radeon RX 7900 XTX, NVIDIA GeForce RTX 3090, NVIDIA GeForce RTX 3090 Ti, AMD Radeon RX 6800. 7 GPUs have plenty of headroom for comfortable inference.

Which devices can run OrcaSAQ 2 27B?

139 devices with unified memory can run OrcaSAQ 2 27B at BF16 (14.3 GB), including AMD Ryzen 7 8845HS (Hawk Point) Laptop, AMD Ryzen AI 7 350 (Krackan Point) Laptop, AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10. Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.