Run GPT-oss 20B on M6

GPT-oss 20B runs at 45 tok/s on a 32 GB M6, a community run reported by github.com (LM Studio, quantization not stated). It needs about 13 GB of unified memory. It is OpenAI's 21B mixture-of-experts model (3.6B active), released under the Apache 2.0 licence.

Speed
community45
tok/s
First Token
—
seconds
RAM Needed
13
GB minimum
Engine
LM Studio
of this figure
Want to run GPT-oss 20B 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 secondcommunity45 tok/s
Provenancecommunity run · github.com
Quantizationquantization not stated
EngineLM Studio
Figure dated2026-10
Minimum RAM13 GB
Parameters21B MoE (3.6B active)
Chip bandwidth170 GB/s

Every figure for GPT-oss 20B on the M6

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 (13 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
community65.6 tok/s4-bitMLX32 GB1komlx.ai12-core GPU. oMLX v0.7.3, MTP/DFlash off; gpt-oss-20b-oQ4, an oQ 4-bit requantization of the MXFP4 release; 10.8 GB peak. 63.8 tok/s at 4k.
community45 tok/squantization not statedLM Studio32 GB—github.com—
Provenance key
sourced
Published figure — vendor or press review
estimated
community
Linked individual run — blog, forum or submission

Speed by context depth: GPT-oss 20B on the M6

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)@4KPrefillTTFT
MLX · 4-bit · 32 GB Mac · omlx.aicommunity65.6 tok/s
1k
community63.8 tok/scommunity1810 tok/s
at 4K
community1.12 s

Setup: run GPT-oss 20B on a Mac

Ollama — install it (brew install ollama or the app from ollama.com), then pull the model by its tag, checked on ollama.com on 2026-11-14:

ollama run gpt-oss:20b

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: Ollama page → · Hugging Face model card →

The 45 tok/s figure above came from LM Studio at quantization not stated; 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.

GPT-oss 20B on other Apple Silicon chips

ChipSpeedQuantEngineBandwidth
M5 Ultracommunity184 tok/sMXFP4MLX1228 GB/s
M5 Maxcommunity141 tok/sMXFP4MLX614 GB/s
M4 Maxestimated135 tok/sMXFP4MLX546 GB/s
M4 Proestimated64 tok/sMXFP4MLX273 GB/s
M4estimated34 tok/sMXFP4MLX120 GB/s

For the estimated figures here, the bandwidth formula’s typical error for small MoEs (≤40B total) is ±15%, against 88 measured runs of 15 models (how it is computed).

About GPT-oss 20B

OpenAI's Apache 2.0 open-weight MoE (Aug 2025): 21B total, 3.6B active, shipped in MXFP4 so it runs in 16 GB. Model card: SWE-bench Verified 53.2, GPQA Diamond 58.6 at medium reasoning. tbreak's review measured 107.5 tok/s on an M5 Max in Ollama, and a community LM Studio run measured ~45 tok/s on a 32 GB Mac mini M6.

Apache 2.0 licence · 21B MoE · 128K context · LLMCheck capability 22/50 · see its leaderboard rank →

Memory by context length

Estimated, not measured: the index’s minimum for GPT-oss 20B (13 GB — the weights at the quant that fits, plus ~2 GB) plus the KV cache for that many tokens, computed from its config.json. 12 full layers + 12 sliding-window layers (window 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 M6 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~0.2 GB~13.1 GB~13.1 GB24 GB Mac18 GB Mac
32K~0.8 GB~13.7 GB~13.4 GB24 GB Macsame
64K (coding-agent context)~1.6 GB~14.5 GB~13.8 GB24 GB Macsame
128K~3.2 GB~16.1 GB~14.6 GB24 GB Macsame

System requirements

Macs with the M6

🛒 Get a Mac that runs GPT-oss 20B

GPT-oss 20B needs about 13 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 GPT-oss 20B stacks up against other models on your specific Mac hardware.

Open Compare Tool Full Leaderboard