Run Nanbeige4.3-33B on M1 Max

Nanbeige4.3-33B runs at 15 tok/s (14.7 as recorded) on a 64 GB M1 Max, a community run reported by omlx.ai (MLX, oQ4e, 1k context). It needs about 17 GB of unified memory. It is Nanbeige's 33.1B dense model, released under the Apache 2.0 licence.

Speed
community15
tok/s
First Token
community15.421
seconds
RAM Needed
17
GB minimum
Engine
MLX
of this figure
Want to run Nanbeige4.3-33B faster — or step up to bigger models? Find your Mac →

The figure

This is a community run: a real run, reported by its source and linked below. The index records it as given and never adjusts it.

MetricValue
Tokens per secondcommunity15 tok/s
Provenancecommunity run · omlx.ai
QuantizationoQ4e
EngineMLX
Time to first token15.421s
Prompt processing (prefill)community63.4 tok/s at 8K context
Context1k
Figure dated2026-11-11
Minimum RAM17 GB
Parameters33.1B
Chip bandwidth400 GB/s

Every figure for Nanbeige4.3-33B on the M1 Max

Each row as the index holds it, with its source. “Mac RAM” is the memory of the Mac a run used, not the model’s requirement (17 GB). The headline figure follows the reference rule: the lower-median measured plain-decode run at 5 bits or fewer (a run whose quant was not stated counts only when no stated-quant run qualifies), else the estimate; a run marked “not used to anchor estimates” is listed here but never becomes the headline.

SpeedQuantEngineMac RAMContextSourceNote
community14.7 tok/soQ4eMLX64 GB1komlx.ai24-core GPU (400 GB/s, the same as the 32-core part). oMLX v0.7.3, MTP/DFlash off; Nanbeige4.3-33B-oQ4e; 18.9 GB peak. 12.4 tok/s at 8k.
Provenance key
sourced
Published figure — vendor or press review
estimated
community
Linked individual run — blog, forum or submission

Speed by context depth: Nanbeige4.3-33B on the M1 Max

Measured figures only, each from its source’s own session; the index does not estimate long-context or prefill speed, so a missing cell is —. Each @ column is plain decode with that many tokens already in the prompt, from the same session as the row’s short-context figure: decode slows as the KV cache it reads grows. Prefill is prompt processing, the tokens per second a runtime reads a prompt at; with long agent contexts it sets how long you wait before the first token.

RunDecode (row)@8KPrefillTTFT
MLX · oQ4e · 64 GB Mac · omlx.aicommunity14.7 tok/s
1k
community12.4 tok/scommunity63.4 tok/s
at 8K
community15.421 s

Setup: run Nanbeige4.3-33B on a Mac

Ollama — the index has not verified an Ollama tag for Nanbeige4.3-33B, so it doesn’t print one (a guessed tag fails with “pull model manifest: file does not exist”). Search the Ollama library →

MLX — pick a build from Hugging Face’s MLX listings, then run it with its repo id:

pip install -U mlx-lm
mlx_lm.generate --model <repo-id> --prompt "Hello"

Downloads: Hugging Face repo → · Hugging Face model card →

The 15 tok/s figure above came from MLX at oQ4e; other runtimes and quants run at different speeds.

Prefer a GUI? LM Studio downloads GGUF and MLX builds in one click. Step by step: install Ollama on a Mac.

Nanbeige4.3-33B on other Apple Silicon chips

ChipSpeedQuantEngineBandwidth
M3 Ultraestimated35 tok/sQ4_K_M—819 GB/s
M5 Maxcommunity25 tok/soQ4eMLX614 GB/s
M4 Maxcommunity25 tok/s4-bitMLX546 GB/s

For the estimated figures here, the bandwidth formula’s typical error for dense models is ±9%, against 75 measured runs of 20 models (how it is computed).

About Nanbeige4.3-33B

Nanbeige's Apache 2.0 dense 33.1B model (24 Oct 2026), 512K context. Artificial Analysis ranks it level with Qwen3.8-27B at the top of its small open models. Vendor-run card: Terminal-Bench 2.1 80.9, SWE-Bench Pro 60.3 (mini-swe-agent, not Scale SEAL), GPQA Diamond 89.5. IQ3_XXS is 14.5 GB and Q4_K_M 20.8 GB, so it targets 24 GB Macs; stock llama.cpp runs it (Llama architecture), and community oMLX runs measured ~25–27 tok/s on an M5 Max at 4-bit.

Apache 2.0 licence · 33.1B · 512K context · LLMCheck capability 46/50 · see its leaderboard rank →

Memory by context length

Estimated, not measured: the index’s minimum for Nanbeige4.3-33B (17 GB — the weights at the quant that fits, plus ~2 GB) plus the KV cache for that many tokens, computed from its config.json. GQA on all 52 layers (8 kv heads x 128). A 16-bit KV cache is the default in llama.cpp, Ollama, LM Studio and mlx-lm; a q8 cache halves the KV part. Coding agents hold 64K tokens or more. Runtimes also add a buffer that grows with context, so leave a few GB spare. macOS lets the GPU use only part of a Mac’s memory by default (figures for an M1 Max Mac on macOS 15, the lower default; macOS 26 allows more): “fits” means within the GPU memory a Mac of that tier can use with the limit raised, a tier tagged raise GPU limit fits only after sudo sysctl iogpu.wired_limit_mb raises it (the tooltip has the exact command), and the last column is the smallest Mac memory size whose default limit holds it (the GPU memory limit).

ContextKV cache (16-bit)RAM neededRAM, q8 KVSmallest tier that fitsWithin the default GPU limit
8K~1.7 GB~17.9 GB~17.4 GB24 GB Mac raise GPU limit32 GB Mac
32K~7.0 GB~23.1 GB~20.1 GB32 GB Mac raise GPU limit (macOS 15)36 GB Mac
64K (coding-agent context)~14.0 GB~30.1 GB~23.5 GB64 GB Mac48 GB Mac
128K~27.9 GB~44.0 GB~30.5 GB64 GB Macsame

System requirements

Macs with the M1 Max

🛒 Get a Mac that runs Nanbeige4.3-33B

Nanbeige4.3-33B needs about 17 GB of unified memory. These Apple Silicon Macs have enough to run it. Apple builds the higher-memory configurations to order; Amazon does not stock them.

As an Amazon Associate, LLMCheck earns from qualifying purchases. These affiliate links cost you nothing extra and help keep the LLMCheck index free.

Compare More Models

See how Nanbeige4.3-33B stacks up against other models on your specific Mac hardware.

Open Compare Tool Full Leaderboard