Skip to page content
SAVRN
Search Plan a project

Open-weight model · Text generation

Qwen3.8-27B-pi-mlx-6Bit

by SolsticeAI SolsticeAI/Qwen3.8-27B-pi-mlx-6Bit

Qwen3.8-27B-pi-mlx-6Bit is an open-weight model for text generation from SolsticeAI, released under Apache License 2.0. It has 26.9B parameters and a 262,144-token context. At 16-bit it needs about 64.6 GB of GPU memory, which fits on 1x MI300X from $1.85 an hour, at the lowest prices in the SAVRN Index. It draws 48 downloads a month.

The Model SolsticeAI/Qwen3.8-27B-pi-mlx-6Bit was converted to MLX format from bytkim/Qwen3.8-27B-pi using mlx-lm version 0.31.2.

Parameters26.9B
Context262,144
Weights21.9 GB
Licenseapache-2.0
AccessOpen weights
Monthly Downloads48

Runs On

What it takes to serve Qwen3.8-27B-pi-mlx-6Bit (26.9B parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.

PrecisionWeightsMemory neededCheapest setupPer hourAlso fits
16-bit 53.8 GB 64.6 GB 1x MI300X (192 GB)
Vultr		 $1.85 1x H100 $1.99 · 1x MI325X $2.00
8-bit 26.9 GB 32.3 GB 1x MI300X (192 GB)
Vultr		 $1.85 1x H100 $1.99 · 1x MI325X $2.00
4-bit 13.4 GB 16.1 GB 1x MI300X (192 GB)
Vultr		 $1.85 1x H100 $1.99 · 1x MI325X $2.00

Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Oct 16, 2026.

Qwen3.8-27B-pi-mlx-6Bit on every accelerator the SAVRN Index prices, at every precision

Model Card

By SolsticeAI, published under apache-2.0, revision 5900d15632f4.

The Model SolsticeAI/Qwen3.8-27B-pi-mlx-6Bit was converted to MLX format from bytkim/Qwen3.8-27B-pi using mlx-lm version 0.31.2.

Read SolsticeAI's full model card

The Model SolsticeAI/Qwen3.8-27B-pi-mlx-6Bit was converted to MLX format from bytkim/Qwen3.8-27B-pi using mlx-lm version 0.31.2.

Use with mlx

pip install mlx-lm

from mlx_lm import load, generate

model, tokenizer = load("SolsticeAI/Qwen3.8-27B-pi-mlx-6Bit")

prompt="hello"

if hasattr(tokenizer, "apply_chat_template") and tokenizer.chat_template is not None:
    messages = [{"role": "user", "content": prompt}]
    prompt = tokenizer.apply_chat_template(
        messages, tokenize=False, add_generation_prompt=True
    )

response = generate(model, tokenizer, prompt=prompt, verbose=True)

Configuration

Architecture
Qwen3_5ForConditionalGeneration
Context length (tokens)
262,144
Layers
64
Hidden size
5,120
Feed-forward size
17,408
Attention heads
24
Key/value heads
4
Head dimension
256
Vocabulary size
248,320
Model type
qwen3_5

Identity and Version

Repository
SolsticeAI/Qwen3.8-27B-pi-mlx-6Bit
Publisher
SolsticeAI
Task
Text generation
Modality
Text
Library
transformers
Parameters
26.9B parameters
Languages
mtp, pi, sft, mlx
Revision
5900d15632f46ea14dfb2e799227b22f45157470
First published
2026-10-11
Last updated
2026-10-15

Files and Weights

13 files, 21.9 GB in total. The weights are 5 files totalling 21.9 GB in safetensors.

Weights5 files · 21.9 GB
Configuration3 files · 194.1 KB
Tokenizer2 files · 20.0 MB
Documentation1 file · 1.3 KB
Other1 file · 51.1 KB
Repository1 file · 1.6 KB
Every file
FileTypeSizeSHA-256
model-00001-of-00005.safetensorsWeights5.4 GB ca2156fd4057
model-00002-of-00005.safetensorsWeights5.3 GB 6ef03d8fa84a
model-00003-of-00005.safetensorsWeights5.3 GB 114264392c5b
model-00004-of-00005.safetensorsWeights4.8 GB fd6b3a5b55fb
model-00005-of-00005.safetensorsWeights1.0 GB 88c510a06aa9
config.jsonConfiguration4.1 KB —
generation_config.jsonConfiguration214 B —
model.safetensors.index.jsonConfiguration189.8 KB —
README.mdDocumentation1.3 KB —
chat_template.jinjaOther51.1 KB —
.gitattributesRepository1.6 KB —
tokenizer.jsonTokenizer20.0 MB 06b9509352d2
tokenizer_config.jsonTokenizer53.4 KB —

License and Download

License
apache-2.0
Access
Open weights, no gate
Download size
21.9 GB
Download from SolsticeAI

Released by SolsticeAI through its official repository on Hugging Face. Read the license.

Built From

  • Derived from bytkim/Qwen3.8-27B-pi

Memory Requirements

PrecisionWeights in memory
As published21.9 GB
16-bit53.8 GB
8-bit26.9 GB
4-bit13.4 GB

Weights only, from the published parameter count; the key-value cache and runtime add to this.

Questions About Qwen3.8-27B-pi-mlx-6Bit

How much GPU memory does Qwen3.8-27B-pi-mlx-6Bit need?

About 64.6 GB at 16-bit and 16.1 GB at 4-bit: the weights (26.9B parameters) plus a working margin. A long context needs more.

What is the cheapest GPU to run Qwen3.8-27B-pi-mlx-6Bit on?

At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.

Can I use Qwen3.8-27B-pi-mlx-6Bit commercially?

Yes. Qwen3.8-27B-pi-mlx-6Bit is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.

What is Qwen3.8-27B-pi-mlx-6Bit's context length?

262,144 tokens, from the maximum position embeddings in its published configuration.

Similar Models

Model · Text generation

Darwin-27B-RSI

FINAL_Bench

Darwin-27B-RSI is Darwin-27B-Opus after Recursive Self-Improvement (RSI): the model was improved using only signal it produced itself. During self-improvement, the model itself (its weights) improves by learning only from its own solutions. No human-written solutions or reasoning traces are used; correctness is checked automatically (agreement across its own samples and code execution). Under an identical evaluation protocol, Darwin-27B-RSI improves over its parent on graduate-level science reasoning — +5.24 points on GPQA Diamond (single sample) and +3.79 points with majority voting — with every gain statistically significant in paired tests. As the reasoning engine of Darwin-27B-JEV on…

Open weights apache-2.0 26.9B parameters 262,144 tokens transformers
View model

Prism ML's ternary Ternary-Bonsai-2-27B build of Qwen/Qwen3.8-27B, repacked for chad, a Claude-Code-style local coding agent for Apple Silicon, with its speculative decoder bundled in. This is chad's default model. Created using Bonsai by Prism ML. with, already quantized. Nothing is built on first run. Every projection of Qwen3.8-27B (a dense qwen35 hybrid: 64 layers, 48 GatedDeltaNet + 16 full attention) is stored in a Hadamard-rotated basis: multiplied by a fixed sign vector and put through a blockwise Walsh-Hadamard transform offline, then quantized to 2-bit affine group-128 whose three levels reproduce the ternary set {−s, 0, +s}. The rotation costs no extra bits and no extra weight…

Open weights apache-2.0 26.9B parameters 262,144 tokens mlx
View model

MLX-optimized version of DavidAU/Qwen3.8-27B-TURBO-Fable-Cold-Fusion-735-882-Heretic-Uncensored-NEO-CODER-MAX-MTP for Apple Silicon inference. Converted from the Q80 GGUF quantization using gguf2mlx. All 866 tensors are dequantized to float16, so this model has identical quality to the original GGUF weights — no precision loss from the conversion. Apache-2.0 (same as the original model)

Open weights apache-2.0 27.3B parameters 262,144 tokens mlx
View model

Benefits high quality CPU inference TQ2 on Llama.cpp and Ollama via QAT - Robotcs, Routing, Coding, Multimedia, Advanced tool calling via JiRackDeltaNetTokenizer - JiRack DeltaNet understand video and images that best for Robotics also A fast and efficient 27B model optimized for CPU inference. Built on a Qwen3.8-style DeltaNet architecture (hybrid attention + SSM), with an updated tokenizer that includes Routing, Media, Vision, Sound, Tool call, and Robotics tags. Ready-to-run GGUF quantizations, and native Ollama support with reasoning disabled by default for fast, direct responses. - JiRack is a cloud-ready model that helps save money on cloud infrastructure. It can be used as an expert…

Open weights mit 27.3B parameters 262,144 tokens
View model

Model · Text generation

Ternary-Bonsai-27B-mlx-2bit

Prism ML

Full 27B-class reasoning in ternary transformer weights — on everyday laptops - \~7.2 GB deployed footprint (down from \~54 GB FP16) — full 27B-class reasoning on a standard laptop or a single GPU - 95% of FP16 intelligence retained: 80.49 average across 15 thinking-mode benchmarks — a higher score than the conventional IQ2XXS build (72.73) at less than two-thirds of its footprint - Retains thinking, reasoning, and agentic behavior deep in the sub-4-bit regime, where conventional low-bit representations collapse: math within two points of full precision (93.40), coding at 85.96, agentic tool use at 74.01 - End-to-end ternary language weights across embeddings, attention projections, MLP…

Open weights apache-2.0 27.4B parameters 262,144 tokens mlx
View model

Qwen3.8-cyber-RedTeam-Surgical-Abliterated (27B) is an unconstrained foundation engine engineered for autonomous cybersecurity agents, vulnerability research, and offensive cyber operations. Built for executing low-level technical directives rather than conversational chatting, it features complete refusal orthogonalization, native multi-step tool calling, and high-throughput linear-attention efficiency: - Quantization Rationale (Native FP8 vs. 4-bit INT4/AWQ): Evaluated across adversary emulation and systems engineering tasks: A standalone orchestration utility is included for automated environment bootstrapping, weight synchronization, and hardware-adaptive server initialization: This…

Open weights apache-2.0 27.4B parameters 262,144 tokens transformers
View model