Home / Models / Nanbeige4.3-33B

How to Run Nanbeige4.3-33B Locally — Setup Guide (October 2026)

Nanbeige4.3-33B is a 33B Apache-2.0 model from Nanbeige LLM Lab (Kanzhun / BOSS Zhipin), built for agentic coding and cowork-style tasks with a 512K-token context window. It uses the plain Llama architecture, so it loads on stock llama.cpp with no fork, and the lab says it is the teacher for an upcoming 3B distill.

Numbers you can trust

We don't print guessed speeds. Every measured number on this page came from TurboLLM's own auto-tuner running Nanbeige4.3-33B on real hardware — labeled as such. When you run it yourself, TurboLLM benchmarks it on your exact GPU and shows a VRAM-fit verdict before you load, and the real measured tokens/sec once it's running.

Does it fit your GPU?

33B dense parameters, Apache-2.0, LlamaForCausalLM: 52 layers, 56 attention heads with 8 KV heads and a 128 head dimension, and a 524,288-token maximum context. The config lists no sliding-window or hybrid layers, so every layer keeps a KV cache that grows with context.

Nanbeige4.3-33B
Nanbeige LLM Lab · GGUF
Dense · Llama architecture · full attention
code
VRAMFit
16 GBIQ2_M (12.19 GB) fits only at a short context · quality drops at 2-bit, so 24 GB is the comfortable floor
24 GBQ4_K_M (20.79 GB) in VRAM with room for a moderate context · 512K maximum context
32 GBQ5_K_M (25.07 GB) or Q6_K (29.02 GB) in VRAM with room for context · 512K maximum context
48 GB+Q8_0 (35.2 GB) fully in VRAM · 512K maximum context
bartowski/Nanbeige_Nanbeige4.3-33B-GGUF
View on Hugging Face →

How to run it

  1. Install TurboLLM

    npx turbollm — no install step, works on Windows, macOS, and Linux. It detects your GPU and auto-provisions a matching llama-server build (no CUDA toolkit, no Python, no compiler).

  2. Get the model

    Paste bartowski/Nanbeige_Nanbeige4.3-33B-GGUF into TurboLLM's in-app Hugging Face search — it lists every quant with a VRAM-fit verdict against your real free VRAM, and downloads are resumable and SHA-256 verified.

  3. Let auto-fit pick the setup

    Flip the auto-fit toggle and TurboLLM chooses the GPU/CPU layer split (and, for a Mixture-of-Experts model, the expert-offload split) for your card — the same manual flag-hunting every other guide walks you through by hand.

  4. Load it

    TurboLLM benchmarks the model on your exact GPU as it loads and shows the real measured tokens/sec — not a number copied from someone else's card.

  5. Use it

    Chat in the built-in UI, or point any OpenAI- or Anthropic-compatible tool (including Claude Code) at TurboLLM's local API — see the API docs.

One honest caveat

The 512K context is the config maximum, not a number you can fill on a consumer card: with full attention on all 52 layers the KV cache grows with every token, so long contexts need a smaller quant, a quantised KV cache, or more memory. The benchmarks on the model card are vendor-run (against Qwen3.8-27B, Muse-Glimmer-30B and Gemma4-31B) and not independently measured, so this page ships without a numbers table. The GGUFs are community conversions by bartowski, and no release date was given by the lab.

FAQ

Will Nanbeige4.3-33B run on my GPU?

Check the VRAM table above for the tier closest to your card. If you're not sure which tier you're in, or want picks tailored to your exact hardware, use What can I run? — enter your VRAM (or Apple unified memory) and it suggests the model, quant, and offload split that fits, the same logic TurboLLM's auto-fit runs when you load a model.

Which quant should I use for Nanbeige4.3-33B?

Pick the highest quant that fits your VRAM with room left over for the KV cache and compute buffer — not just the model weights. As a starting point, Q4_K_M is usually the sweet spot (small quality loss for a large size drop); step up to Q5_K_M/Q6_K/Q8_0 if you have headroom, or down to Q3_K_M/an IQ quant if you don't. TurboLLM shows a VRAM-fit verdict for every quant against your card's real free VRAM before you download — see Quantization explained for what the letters and numbers actually mean, or llama.cpp flags that actually matter for the other settings worth knowing.

Why does Nanbeige4.3-33B suddenly get slow or stutter?

The usual cause is silent VRAM spill: once a model's weights plus its KV cache stop fitting in VRAM, the driver quietly offloads the overflow to system RAM over PCIe, and that path runs 5–10× slower with no error message — just a sudden slowdown. TurboLLM's auto-fit sizes the quant and GPU/CPU split against your card's real free VRAM (including KV-cache growth at your chosen context) specifically to avoid this, and shows a VRAM-fit verdict before you load rather than after it's already crawling. Full mechanics in VRAM spill explained.

Run it now

$ npx turbollm

One command detects your GPU, provisions the right engine, and opens the UI. New here? Start with Install & first run and Quantization explained. On different hardware, see RTX 3090, RTX 4090 or RTX 5080. Also see how to run it or how to run it.