---
license: other
base_model: stepfun-ai/Step-3.5-Flash
tags:
  - gguf
  - quantized
  - apex
  - moe
  - mixture-of-experts
  - step
---





<!-- apex-banner-v2 -->
<div style="background-color: #f59e0b; color: white; padding: 20px; border-radius: 10px; text-align: center; margin: 20px 0;">
<h2 style="color: white; margin: 0 0 10px 0;">⚡ Each donation = another big MoE quantized</h2>
<p style="font-size: 18px; margin: 0 0 15px 0;">I host <b>25+ free APEX MoE quantizations</b> as independent research. My only local hardware is an <b>NVIDIA DGX Spark</b> (122 GB unified memory) — enough for ~30-50B-class MoEs, but <b>bigger ones (200B+) require rented compute</b> on H100/H200/Blackwell, typically $20-100 per quant.<br>If APEX quants are useful to you, your support directly funds those bigger runs.</p>
<p style="font-size: 20px; margin: 0;">
<a href="https://www.patreon.com/cw/mudler" style="color: white; text-decoration: underline;">🎉 Patreon (Monthly)</a> &nbsp;|&nbsp;
<a href="https://www.buymeacoffee.com/mudler" style="color: white; text-decoration: underline;">☕ Buy Me a Coffee</a> &nbsp;|&nbsp;
<a href="https://github.com/sponsors/mudler" style="color: white; text-decoration: underline;">⭐ GitHub Sponsors</a>
</p>
<p style="font-size: 14px; margin: 10px 0 0 0; opacity: 0.9;">💚 Big thanks to Hugging Face for generously donating additional storage — much appreciated.</p>
</div>

# Step-3.5-Flash APEX GGUF

**APEX (Adaptive Precision for EXpert Models)** quantizations of [Step-3.5-Flash](https://huggingface.co/stepfun-ai/Step-3.5-Flash).

**Brought to you by the [LocalAI](https://github.com/mudler/LocalAI) team** | [APEX Project](https://github.com/mudler/apex-quant) | [Technical Report](https://github.com/mudler/apex-quant/blob/main/paper/APEX_Technical_Report.pdf)

## Benchmark Results

Benchmarks coming soon. For reference APEX benchmarks on the Qwen3.5-35B-A3B architecture, see [mudler/Qwen3.5-35B-A3B-APEX-GGUF](https://huggingface.co/mudler/Qwen3.5-35B-A3B-APEX-GGUF).

## What is APEX?

APEX is a quantization strategy for Mixture-of-Experts (MoE) models. It classifies tensors by role (routed expert, shared expert, attention) and applies a layer-wise precision gradient -- edge layers get higher precision, middle layers get more aggressive compression. I-variants use diverse imatrix calibration (chat, code, reasoning, tool-calling, agentic traces, Wikipedia).

See the [APEX project](https://github.com/mudler/apex-quant) for full details, technical report, and scripts.

## Architecture

- **Model**: Step-3.5-Flash (Step3p5)
- **Layers**: 45 (3 dense + 42 MoE)
- **Experts**: 288 routed + 1 shared (top-8 active per token)
- **Total Parameters**: ~196B
- **Active Parameters**: ~11B per token
- **Context**: 256K tokens (MTP 4-token speculative head)
- **APEX Config**: 5+5 symmetric edge gradient across 45 layers

## Run with LocalAI

```bash
local-ai run mudler/Step-3.5-Flash-APEX-GGUF@Step-3.5-Flash-APEX-I-Balanced.gguf
```

## Credits

APEX is brought to you by the [LocalAI](https://github.com/mudler/LocalAI) team. Developed through human-driven, AI-assisted research. Built on [llama.cpp](https://github.com/ggerganov/llama.cpp).