The best local LLM for a Mac Studio M2 Max (64 GB) is KAT-Coder-V2.5 at ~84 tok/s (estimated). Its 64 GB fits 77 of the 130 models in the LLMCheck index, up to 124B-class. For everyday chat and coding, KAT-Coder-V2.5 is the sweet spot.
Considering this Mac?
Mac Studio M5 Max (64GB): build-to-order from Apple · Mac Studio M2 Max on Amazon Renewed → · compare every Mac for local AI →
Ranked by LLMCheck suitability (capability balanced against speed on the M2 Max); the top pick is the best-ranked model that reaches 15 tok/s here, has a standard Mac runtime and fits the GPU memory limit (raised if need be), and slower models are tagged slow. Click a model for its full figures and setup. Every speed carries its provenance mark: most are index estimates from the memory-bandwidth model, a filled square is a published figure and a hollow circle a community run — submit a real run →
| # | Model | Size | License | Speed | 64K context | Capability |
|---|---|---|---|---|---|---|
| 1 | KAT-Coder-V2.5 | 35B MoE | Apache 2.0 | estimated84 tok/s | agent-ready | 39/50 |
| 2 | Qwen 3.6-35B-A3B | 35B MoE | Apache 2.0 | estimated92 tok/s | agent-ready | 38/50 |
| 3 | Qwen3.8-27B | 27.8B | Apache 2.0 | estimated21 tok/s | agent-ready | 46/50 |
| 4 | Nanbeige4.3-33B | 33.1B | Apache 2.0 | estimated15 tok/s | agent-ready | 46/50 |
| 5 | Recursive Flint | 39.8B MoE | Apache 2.0 | estimated89 tok/s | agent-ready | 37/50 |
| 6 | Ling-3.0-flash-VL raise GPU limit | 124B MoE | MIT | estimated46 tok/s | 42/50 | |
| 7 | Qwen 3.6-27B | 27B | Apache 2.0 | estimated20 tok/s | agent-ready | 44/50 |
| 8 | Nemotron 3.5 Lightning | 30B MoE | OpenMDW | estimated85 tok/s | agent-ready | 36/50 |
| 9 | North Mini Code 1.0 | 30.5B MoE | Apache 2.0 | estimated82 tok/s | agent-ready | 36/50 |
| 10 | LFM2.5-24B-A2B | 24B MoE | LFM Open | estimated90 tok/s | agent-ready | 35/50 |
| 11 | GLM-4.7-Flash | 31B MoE | MIT | estimated86 tok/s | agent-ready | 35/50 |
| 12 | Gemma 4 26B-A4B | 26B MoE | Apache 2.0 | estimated86 tok/s | agent-ready | 35/50 |
Showing the top 12 of 77 models that fit in 64 GB. See the full leaderboard or all benchmarks.
Context: “fits” means the weights at a short prompt. Coding agents (Claude Code, OpenCode, Codex CLI) hold 64K tokens or more, and every token adds KV cache: agent-ready marks a model whose weights plus a 64K-token 16-bit KV cache still fit in the 56 GB this Mac’s GPU can use with the limit raised (60 of the 70 models here with a KV figure). That is an estimate from each model’s config.json, not a measurement; a q8 KV cache halves the cache part, and — means the index has no KV figure for the model yet. The GPU memory limit below applies on top.
GPU memory limit: macOS lets the GPU use about 48 GB of this Mac’s 64 GB by default on macOS 15 and earlier, and about 51.8 GB on macOS 26 and later. Of the 77 models that fit, 6 need more than that at a short prompt: 3 need a raised limit and are tagged raise GPU limit; 3 need more than even the raised limit and are tagged over GPU limit. sudo sysctl iogpu.wired_limit_mb=57344 raises it to 56 GB, leaving 8 GB for macOS, until the Mac restarts (sudo sysctl iogpu.wired_limit_mb=0 restores the default). How the limit is read →
The index has no verified Ollama tag for KAT-Coder-V2.5, so start from its listing and copy the exact name there: Search the Ollama library → · model card →
GGUF builds run in LM Studio or llama.cpp; MLX builds run in LM Studio or mlx-lm.
Prefer a GUI? LM Studio gives you a one-click download and chat window. For step-by-step help see our Ollama install guide, or open the KAT-Coder-V2.5 on M2 Max page for its figures and setup.
Twice as fast: no Mac in the index runs KAT-Coder-V2.5 at twice this Mac’s estimated~84 tok/s; the fastest figure is the Mac Studio M5 Ultra (96 GB) at estimated~120 tok/s, $5,499 new (Apple list price, checked 2026-11-14): build-to-order from Apple (Amazon does not stock this configuration).
For less: the cheapest Mac that still runs KAT-Coder-V2.5 at 15+ tok/s is the MacBook Air M2 (24 GB) at estimated~31 tok/s, about $850 refurbished or used: search Amazon Renewed listings →. New: the Mac mini M6 (24 GB) at estimated~47 tok/s, $1,099 new (Apple list price, checked 2026-11-14): Mac mini M6 (24GB) on Amazon →.
Speeds are each Mac’s own figure from the index, with its provenance mark. Prices come from data/macs.json: Apple’s list price for a new Mac (dated), an approximate street price for a used or refurbished one. Amazon stock checked 2026-11-18. Amazon links are affiliate links: as an Amazon Associate, LLMCheck earns from qualifying purchases.
Apple no longer sells the Mac Studio M2 Max (64 GB) new. A current Mac Studio M5 Max (64 GB) runs the same models with 614 GB/s of memory bandwidth instead of 400 GB/s, so they generate faster. The Mac Studio M5 Max (64 GB) is build-to-order from Apple; Amazon does not stock that configuration.
As an Amazon Associate, LLMCheck earns from qualifying purchases. Affiliate links cost you nothing extra and never influence our rankings.
LLMCheck. (2026). Best Local LLMs for the Mac Studio M2 Max (64 GB). llmcheck.net. https://llmcheck.net/best-llm/mac-studio-m2-max-64gb/ (updated ). CC BY 4.0.
KAT-Coder-V2.5 (35B MoE, Apache 2.0) is the best all-round pick at ~84 tok/s (estimated) on the Mac Studio M2 Max. If you want maximum speed, LFM3-1.7B hits ~255 tok/s (estimated); for maximum capability, Qwen3.8-27B still fits in 64 GB.
About 77 of the 130 models in the LLMCheck index fit in 64 GB of unified memory, from compact models up to Ling-3.0-flash-VL (124B MoE).
Yes. A dense 70B such as Llama 3.3 70B needs about 44 GB at 4-bit, which fits in 64 GB with room for context; the index puts it at ~8 tok/s (estimated) on this Mac.
64 GB is enough for dense models up to about 72B at 4-bit (the largest in the index that fits is Qwen 2.5 72B), and for mixture-of-experts models such as Ling-3.0-flash-VL (~46 tok/s (estimated)). As a rule of thumb a 4-bit model needs about 0.57 GB per billion parameters plus ~2 GB, and macOS needs its own share of memory too. Because Apple Silicon uses unified memory, the same pool serves as system RAM and GPU memory.