Run MiniCPM5-2B on M5 Max

The LLMCheck index estimates MiniCPM5-2B at ~225 tok/s on an M5 Max at Q4_K_M, a figure from its published bandwidth model, not a measured run. It needs about 3 GB of unified memory. It is OpenBMB's 2.52B dense model, released under the Apache 2.0 licence.

Speed
estimated~225
tok/s
First Token
—
seconds
RAM Needed
3
GB minimum
Engine
llama.cpp
of this figure
Want to run MiniCPM5-2B faster — or step up to bigger models? Find your Mac →

The figure

This is an LLMCheck index estimate from its published two-term bandwidth model, not a measured run; the index labels it so everywhere it appears. The bandwidth formula’s typical error for dense models is ±9%, against 75 measured runs of 20 models (how it is computed). Own this Mac? Submit a real run →

MetricValue
Tokens per secondestimated~225 tok/s
Provenanceestimate · how estimates are made
QuantizationQ4_K_M
Enginellama.cpp
Figure dated2026-11
Minimum RAM3 GB
Parameters2.52B
Chip bandwidth614 GB/s

Every figure for MiniCPM5-2B on the M5 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 (3 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
community83.8 tok/s8-bitMLX64 GB1komlx.ai40-core GPU. oMLX v0.7.0.dev4, MTP/DFlash off; MiniCPM5-2B-8bit; 3.3 GB peak. The session speeds up with context (90.6 tok/s at 4k, 101.6 at 8k) and its 1k figure sits below this model's 4-bit runs on the half-bandwidth M5 Pro, so it is not a steady decode rate and does not anchor.; not used to anchor estimates
estimated225 tok/sQ4_K_Mllama.cpp48 GB—LLMCheck estimator—
Provenance key
sourced
Published figure — vendor or press review
estimated
community
Linked individual run — blog, forum or submission

Speed by context depth: MiniCPM5-2B on the M5 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)@4K@8K@16K@32KPrefillTTFT
MLX · 8-bit · 64 GB Mac · omlx.aicommunity83.8 tok/s
1k
community90.6 tok/scommunity101.6 tok/scommunity78.7 tok/scommunity63.8 tok/scommunity2858 tok/s
at 8K
community0.358 s

Setup: run MiniCPM5-2B on a Mac

Ollama — the index has not verified an Ollama tag for MiniCPM5-2B, 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 →

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

MiniCPM5-2B on other Apple Silicon chips

ChipSpeedQuantEngineBandwidth
M5 Procommunity144 tok/s4-bitMLX307 GB/s

About MiniCPM5-2B

OpenBMB's Apache 2.0 2.5B dense model (6 Sep 2026), 128K context. Artificial Analysis ranks it among the two strongest open models of 4B or fewer parameters. Vendor card: GPQA Diamond 70.2, MMLU-Pro 70.8. Q4_K_M is 1.6 GB; stock llama.cpp and the official MLX 4-bit build run it, and SparkLLM's card chart lists 147 tok/s on an M5 Pro (mlx-lm 0.32.2, 4-bit).

Apache 2.0 licence · 2.52B · 128K context · LLMCheck capability 21/50 · see its leaderboard rank →

Memory by context length

Estimated, not measured: the index’s minimum for MiniCPM5-2B (3 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 42 layers (2 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 M5 Max Mac, which runs macOS 26 or later): “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~0.4 GB~3.2 GB~3.1 GB8 GB Macsame
32K~1.4 GB~4.2 GB~3.6 GB8 GB Macsame
64K (coding-agent context)~2.8 GB~5.6 GB~4.3 GB16 GB Macsame
128K~5.6 GB~8.5 GB~5.7 GB16 GB Macsame

System requirements

Macs with the M5 Max

🛒 Get a Mac that runs MiniCPM5-2B

MiniCPM5-2B needs about 3 GB of unified memory. These Apple Silicon Macs have enough to run it.

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 MiniCPM5-2B stacks up against other models on your specific Mac hardware.

Open Compare Tool Full Leaderboard