---
title: "Best Local LLMs for MacBook Air M2 24GB (2026)"
canonical: https://llmcheck.net/best-llm/macbook-air-m2-24gb/
markdown: https://llmcheck.net/md/best-llm/macbook-air-m2-24gb.md
page_type: best-llm-mac
publisher: "LLMCheck (llmcheck.net)"
data_version: 2026-11-18
catalog_verified_through: 2026-11-18
updated: 2026-11-19
license: "CC BY 4.0"
license_url: https://creativecommons.org/licenses/by/4.0/
cite: "LLMCheck. (2026). Best Local LLMs for the MacBook Air M2 (24 GB). llmcheck.net. https://llmcheck.net/best-llm/macbook-air-m2-24gb/ (updated 2026-11-19). CC BY 4.0."
provenance:
  sourced: "Published figure — vendor or press review"
  community: "Linked individual run — blog, forum or submission"
  estimated: "Derived from published bandwidth model (https://llmcheck.net/methodology#estimation)"
---

# Best Local LLMs for the MacBook Air M2 (24 GB)

Markdown twin of https://llmcheck.net/best-llm/macbook-air-m2-24gb/, generated from the LLMCheck dataset (data version 2026-11-18) with the same figures as the page. Every speed names its provenance: **sourced** (a published vendor or press figure), **community** (one person's linked run) or **estimated** (the bandwidth model). LLMCheck is an index and runs no tests of its own. Licence CC BY 4.0: credit "LLMCheck (llmcheck.net)" and link the page.

According to the LLMCheck index (18 November 2026), the best local LLM for a MacBook Air M2 (24 GB) is **Qwen 3.6-35B-A3B** at ~43 tok/s estimated: the best-ranked model that reaches 15 tok/s on this Mac and has a standard Mac runtime. 63 of the 130 models in the index fit in 24 GB; 39 of them reach 15 tok/s here.

## This Mac

| Spec | Value |
| --- | --- |
| Unified memory | 24 GB |
| Memory bandwidth | 100 GB/s |
| Models that fit | 63 of 130 |
| Top speed | 106 tok/s (estimated) — Ling-3.0-tiny |
| Status | no longer sold new (used / refurbished) |
| Price | about $850 refurbished or used |

## Top 3 picks for the MacBook Air M2 (24 GB)

- **Best overall: Qwen 3.6-35B-A3B** — 43 tok/s (estimated); 35B MoE, Apache 2.0, capability 38/50; raise GPU limit
- **Fastest: Ling-3.0-tiny** — 106 tok/s (estimated); 7.9B MoE, MIT, capability 15/50; agent-ready at 64K
- **Most capable (slow): Qwen3.8-27B** — 5 tok/s (estimated); 27.8B, Apache 2.0, capability 46/50; raise GPU limit

## Every model ranked for a MacBook Air M2 (24 GB) (top 12 of 63)

Ranked by LLMCheck suitability: capability plus min(15, tok/s × 0.12) on the M2 (100 GB/s). Speeds are per-config figures for this Mac; a measured figure is the chip's run.

| # | Model | Size | License | tok/s | Provenance | Quant | 64K context | Capability | Note |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| 1 | Qwen 3.6-35B-A3B | 35B MoE | Apache 2.0 | 43 | estimated | Q4_K_M | needs ~26 GB | 38/50 | raise GPU limit: needs about 24 GB by the catalog's build; macOS lets the GPU use ~16 GB of 24 GB by default. A published run's build peaked at 19.4 GB, inside the 20 GB that sudo sysctl iogpu.wired_limit_mb=20480 allows until the next restart — at a short context only |
| 2 | Qwen3.8-27B | 27.8B | Apache 2.0 | 5 | estimated | Q4_K_M | needs ~24 GB | 46/50 | slow (under 15 tok/s); raise GPU limit: needs about 19 GB; macOS lets the GPU use ~16 GB of 24 GB by default. sudo sysctl iogpu.wired_limit_mb=20480 raises it to 20 GB until the next restart |
| 3 | Nanbeige4.3-33B | 33.1B | Apache 2.0 | 4 | estimated | Q4_K_M | needs ~31 GB | 46/50 | slow (under 15 tok/s); raise GPU limit (macOS 15): needs about 17 GB; macOS lets the GPU use ~16 GB of 24 GB by default on macOS 15 and earlier (~17.8 GB on macOS 26 and later, enough). sudo sysctl iogpu.wired_limit_mb=20480 raises it to 20 GB until the next restart |
| 4 | Qwen 3.6-27B | 27B | Apache 2.0 | 5 | estimated | Q4_K_M | needs ~23 GB | 44/50 | slow (under 15 tok/s); raise GPU limit: needs about 18 GB; macOS lets the GPU use ~16 GB of 24 GB by default. sudo sysctl iogpu.wired_limit_mb=20480 raises it to 20 GB until the next restart |
| 5 | Recursive Flint | 39.8B MoE | Apache 2.0 | 49 | estimated | 3.8 bpw | needs ~22 GB | 37/50 | raise GPU limit: needs about 20 GB; macOS lets the GPU use ~16 GB of 24 GB by default. sudo sysctl iogpu.wired_limit_mb=20480 raises it to 20 GB until the next restart |
| 6 | KAT-Coder-V2.5 | 35B MoE | Apache 2.0 | 31 | estimated | Q4_K_M | needs ~26 GB | 39/50 | over GPU limit: needs about 24 GB; macOS lets the GPU use ~16 GB of 24 GB by default, and a safe raise reaches 20 GB (leaving 4 GB for macOS): expect heavy swapping, a CPU split or an out-of-memory error; a smaller quant or a shorter context may fit |
| 7 | Muse Glimmer 30B | 30B | Apache 2.0 | 4 | estimated | Q4_K_M | agent-ready (~19 GB of 20 GB usable) | 42/50 | slow (under 15 tok/s); raise GPU limit: needs about 18 GB; macOS lets the GPU use ~16 GB of 24 GB by default. sudo sysctl iogpu.wired_limit_mb=20480 raises it to 20 GB until the next restart |
| 8 | LFM2.5-24B-A2B | 24B MoE | LFM Open | 52 | estimated | Q4_K_M | agent-ready (~19 GB of 20 GB usable) | 35/50 | raise GPU limit (macOS 15): needs about 17 GB; macOS lets the GPU use ~16 GB of 24 GB by default on macOS 15 and earlier (~17.8 GB on macOS 26 and later, enough). sudo sysctl iogpu.wired_limit_mb=20480 raises it to 20 GB until the next restart |
| 9 | Gemma 4 31B | 31B | Apache 2.0 | 5 | estimated | Q4_K_M | needs ~30 GB | 40/50 | slow (under 15 tok/s); raise GPU limit: needs about 24 GB by the catalog's build; macOS lets the GPU use ~16 GB of 24 GB by default. A published run's build peaked at 17.8 GB, inside the 20 GB that sudo sysctl iogpu.wired_limit_mb=20480 allows until the next restart — at a short context only |
| 10 | MiniCPM-V 4.7 35B-A3B | 35.2B MoE | Apache 2.0 | 38 | estimated | Q4_K_M | needs ~25 GB | 35/50 | over GPU limit: needs about 23 GB; macOS lets the GPU use ~16 GB of 24 GB by default, and a safe raise reaches 20 GB (leaving 4 GB for macOS): expect heavy swapping, a CPU split or an out-of-memory error; a smaller quant or a shorter context may fit |
| 11 | Nemotron 3.5 Lightning | 30B MoE | OpenMDW | 29 | estimated | Q4_K_M | agent-ready (~19 GB of 20 GB usable) | 36/50 | raise GPU limit: needs about 18 GB; macOS lets the GPU use ~16 GB of 24 GB by default. sudo sysctl iogpu.wired_limit_mb=20480 raises it to 20 GB until the next restart |
| 12 | Gemma 4 26B-A4B | 26B MoE | Apache 2.0 | 34 | estimated | Q4_K_M | agent-ready (~20 GB of 20 GB usable) | 35/50 | raise GPU limit: needs about 18 GB; macOS lets the GPU use ~16 GB of 24 GB by default. sudo sysctl iogpu.wired_limit_mb=20480 raises it to 20 GB until the next restart |

Provenance: **sourced** = Published figure — vendor or press review; **community** = Linked individual run — blog, forum or submission; **estimated** = Derived from published bandwidth model (https://llmcheck.net/methodology#estimation).

## Context

A model is agent-ready at 64K when its weights plus a 65,536-token 16-bit KV cache fit in the 20 GB this Mac's GPU can use with the limit raised (33 of the 58 models here with a KV figure); an estimate from each model's config.json, not a measurement. A q8 KV cache halves the cache part. **GPU memory limit:** macOS lets the GPU use about 16 GB of this Mac’s 24 GB by default on macOS 15 and earlier, and about 17.8 GB on macOS 26 and later. Of the 63 models that fit, 25 need more than that at a short prompt: 17 need a raised limit and are tagged raise GPU limit (Qwen 3.6-35B-A3B and Gemma 4 31B only past the safe setting below, as measured runs on this much memory show); 8 need more than even the raised limit and are tagged over GPU limit. `sudo sysctl iogpu.wired_limit_mb=20480` raises it to 20 GB, leaving 4 GB for macOS, until the Mac restarts (`sudo sysctl iogpu.wired_limit_mb=0` restores the default). [How the limit is read →](https://llmcheck.net/methodology#gpu-limit)

Estimated figures come from the bandwidth formula; against 163 measured runs its typical error is ±9% for dense models and ±15% for small MoEs (≤40B total). [How the error is computed →](https://llmcheck.net/methodology#typical-error)

## Quick start: Qwen 3.6-35B-A3B on the MacBook Air M2

Ollama tag, verified on ollama.com 2026-11-14: `ollama run qwen3.6:35b-a3b`

Model card: https://huggingface.co/Qwen/Qwen3.6-35B-A3B

## Related

- MacBook Air M2 (16 GB): 38 models fit — https://llmcheck.net/md/best-llm/macbook-air-m2-16gb.md
- Best local LLM for every 24 GB Mac — https://llmcheck.net/md/best-llm/by-ram.md
- Every Mac — https://llmcheck.net/md/best-llm.md
- Full leaderboard — https://llmcheck.net/md/leaderboard.md


## Cite this page

LLMCheck. (2026). Best Local LLMs for the MacBook Air M2 (24 GB). llmcheck.net. https://llmcheck.net/best-llm/macbook-air-m2-24gb/ (updated 2026-11-19). CC BY 4.0.

How to cite: https://llmcheck.net/cite · Open data (CSV, JSON): https://llmcheck.net/data/ · Every Markdown mirror: https://llmcheck.net/md/index.md
