Skip to content

What can the M3 Ultra 512 GB run?

What fits on the M3 Ultra 512 GB
Text models: 69Visual models: 0Strongest first
Memory
512 GB unified
Bandwidth
819 GB/s
Maker
Apple

Open the calculator on this machineApple M3 Ultra in our inventory

Text models that run on it

Text models that run on the M3 Ultra 512 GB, best first
ModelBest fitContextSpeedStrongest at
MiMo V2.6 Pro2-bit128K25 tok/sChat, Coding, Agents, Docs
DeepSeek V4.1 Flash3-bit1M47 tok/sChat, Coding, Agents, Docs
GLM-5.33-bit216K21 tok/sChat, Coding, Agents, Docs
GLM-5.3 FlashFP81M24 tok/sChat, Coding, Agents, Docs
Qwen3.8 Flash-Next8-bit256K53 tok/sChat, Coding, Agents
Muse Spark Open 816B A42B2-bit1M26 tok/sChat, Docs
MiMo V2.6 FlashFP4 + FP81M37 tok/sCoding, Agents
MiniMax M36-bit320K23 tok/sDocs
DeepSeek V4 FlashFP4 + FP81M24 tok/sDocs
DeepSeek V4 Pro1.6-bit1M36 tok/sChat, Coding, Agents, Docs
Kimi K2.62-bit80K32 tok/sChat, Coding, Agents, Docs
Hy38-bit224K19 tok/sChat, Coding, Agents, Docs
Nemotron 3 Ultra 550B A55B4-bit256K13 tok/sChat, Coding, Docs
Qwen3.5 397B A17B6-bit256K29 tok/sAgents, Docs
MiniMax M2.7FP8200K36 tok/sCoding, Agents
DeepSeek V3.23-bit160K24 tok/sChat, Coding, Agents, Docs
Mistral Medium 3.5 128BFP8256K3.9 tok/sChat, Coding, Agents, Docs
Qwen3.5 122B A10B8-bit256K37 tok/sChat, Coding, Agents, Docs
Nemotron 3 Super 120B A12B8-bit256K27 tok/sChat, Coding, Agents, Docs
Mistral Small 4 119BFP8256K57 tok/sChat, Docs
Inkling2-bit4K26 tok/sText and images
DeepSeek R13-bit160K22 tok/sNot rated
DeepSeek V33-bit160K22 tok/sNot rated
GLM 53-bit198K20 tok/sNot rated
GLM 4.5 Air8-bit128K28 tok/sNot rated
Inkling Small8-bit4K35 tok/sText and images
MiniMax M2.5FP8192K33 tok/sNot rated
GLM 5.13-bit198K20 tok/sNot rated
Laguna M.18-bit256K18 tok/sNot rated
Laguna S 2.18-bit1M42 tok/sNot rated
Qwen3 VL 235B A22B Instruct8-bit256K18 tok/sText and images
GLM 5.23-bit216K20 tok/sNot rated
Qwen3 235B A22B8-bit40K18 tok/sNot rated
Ornith 1.5 122B8-bit256K31 tok/sNot rated
Kimi K2 Instruct2-bit80K26 tok/sNot rated
Ornith 1.5 397B6-bit256K22 tok/sNot rated
Ornith 1.0 397B6-bit256K28 tok/sNot rated
Step 3.5 Flash8-bit256K40 tok/sNot rated
DeepSeek V3 03243-bit160K22 tok/sNot rated
DeepSeek R1 05283-bit160K22 tok/sNot rated
GLM 4.58-bit24K12 tok/sNot rated
Llama 4 Scout 17B 16E Instruct8-bit1.3M23 tok/sText and images
MiniMax M2FP8192K33 tok/sNot rated
MiMo V2.5FP81M25 tok/sNot rated
Solar Open 100B8-bit128K32 tok/sNot rated
GLM 5.3 Flash BF168-bit1M23 tok/sText and images
Qwen3 VL 235B A22B Thinking8-bit256K18 tok/sText and images
Qwen3 235B A22B Instruct 25078-bit256K18 tok/sNot rated
NVIDIA Nemotron 3.5 Super VL 120B A12B BF168-bit1M27 tok/sText and images
Hy3 Preview8-bit216K19 tok/sNot rated
Qwen3 Coder 480B A35B Instruct5-bit184K18 tok/sCoding
GLM 4.78-bit24K12 tok/sNot rated
GLM 5.3 BF163-bit216K20 tok/sNot rated
GLM 4.68-bit28K12 tok/sNot rated
DeepSeek V2 Chat8-bit160K20 tok/sNot rated
MiMo V2 FlashFP8256K26 tok/sNot rated
Le Triomphant ECE TW38-bit32K6.7 tok/sNot rated
TW3 JRGL V28-bit32K6.7 tok/sNot rated
Command A Plus 05 2026 Bf168-bit195K18 tok/sText and images
GLM 4.5V8-bit64K29 tok/sText and images

Best fit is the highest quality, up to 8-bit, that leaves the use case’s context. Ranked by our use-case scores, then by Hugging Face downloads. Speeds are rough, plus or minus 25%. The top 60 of 69 are shown.

Only just too big

These need up to a quarter more memory than the M3 Ultra 512 GB has. Spilling into system RAM or a lower precision can get them running, slowly.

Questions

What is the best LLM for the M3 Ultra 512 GB?

Of the open models we track, the strongest that run well on the M3 Ultra 512 GB are MiMo V2.6 Pro, DeepSeek V4.1 Flash, GLM-5.3. 69 text models fit it in all.

Calculated from published specs, not benchmarked. All machines