All models
Alibaba Qwen
Qwen3 Coder Next
Qwen's coding model for agent tools like opencode and aider. 80B total with 3B active, so it stays fast on long coding sessions.
moecodeagents
Overview
Qwen3 Coder Next is a mixture-of-experts coding model with 80B total and 3B activated parameters, and a 262,144 token context. It is aimed at coding agents and repository-scale work.
Hardware fit
Q4_K_M is 48.5 GB and fits an 80 GB GPU with plenty of context. Q8_0 at 84.8 GB needs a 141 GB H200 or two GPUs.
| Quantization | Weights |
|---|---|
| Q4_K_M | 48.5 GB |
| Q6_K | 65.6 GB |
| Q8_0 | 84.8 GB |
Sizes are the GGUF file sizes from unsloth/Qwen3-Coder-Next-GGUF. Budget extra VRAM for the KV cache, which grows with context size and parallel requests.
Deploying on Loadcore
- In the console, open Models and search for
unsloth/Qwen3-Coder-Next-GGUF. - Pick the quantization that fits the target server and start the download.
- Start an instance with the context size you need and GPU layers set to offload the full model.
- Point your tools at
http://your-server:9090/v1with your personal key.
Recommended settings
| Parameter | Value |
|---|---|
| temperature | 1.0 |
| top_p | 0.95 |
| top_k | 40 |
| repeat_penalty | 1.0 |
| min_p | 0.0 |
Values below are Qwen's recommended sampling for this model.
Source: Qwen/Qwen3-Coder-Next.