Intel · Core Ultra 9 285H · Laptop

Best AI Models for Intel Core Ultra 9 285H (Arrow Lake) Laptop (32 GB)

Runs 35B MoE models well; 9B, 14B and 27–31B fit but are slow (9, 6 and 3 tok/s), and 70B doesn't fit in 24 GB.

32 GB unified24 GB for models119.5 GB/s39 TFLOPS FP1645 W2025

What LLMs can the Intel Core Ultra 9 285H (Arrow Lake) Laptop run?

#	Model	Speed	Quant · Memory
1	Qwen3.6 35B A3BAlibabaHow to run → Q4_K_M · 22.0 GB · 17.7 t/s	17.7 tok/s	Q4_K_M · 22.0 GB
2	Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 19.2 t/s	19.2 tok/s	Q4_K_M · 16.5 GB
3	Qwen3 30B A3B Thinking 2507Alibaba Q4_K_M · 19.4 GB · 22.7 t/s	22.7 tok/s	Q4_K_M · 19.4 GB
4	GPT OSS 20BOpenAI Q4_K_M · 13.1 GB · 20.8 t/s	20.8 tok/s	Q4_K_M · 13.1 GB
5	NVIDIA Nemotron 3.5 Lightning 30B A3B BF16NVIDIA Q4_K_M · 19.3 GB · 28.9 t/s	28.9 tok/s	Q4_K_M · 19.3 GB

#	Model	Speed	Coding Score	Quant · Memory
1	Qwen3.6 35B A3BAlibabaHow to run → Q4_K_M · 22.0 GB · 17.7 t/s	17.7 tok/s		Q4_K_M · 22.0 GB
2	Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 19.2 t/s	19.2 tok/s		Q4_K_M · 16.5 GB
3	GPT OSS 20BOpenAI Q4_K_M · 13.1 GB · 20.8 t/s	20.8 tok/s		Q4_K_M · 13.1 GB
4	Qwen3 30B A3B Thinking 2507Alibaba Q4_K_M · 19.4 GB · 22.7 t/s	22.7 tok/s	overall	Q4_K_M · 19.4 GB
5	NVIDIA Nemotron 3.5 Lightning 30B A3B BF16NVIDIA Q4_K_M · 19.3 GB · 28.9 t/s	28.9 tok/s		Q4_K_M · 19.3 GB

#	Model	Speed	Quant · Memory
1	Gemma 4 26B A4B ITGoogleHow to run → Q4_K_M · 16.5 GB · 19.2 t/s	19.2 tok/s	Q4_K_M · 16.5 GB
2	Qwen3.5 4BAlibaba Q4_K_M · 3.4 GB · 18.3 t/s	18.3 tok/s	Q4_K_M · 3.4 GB
3	GPT OSS 20BOpenAI Q4_K_M · 13.1 GB · 20.8 t/s	20.8 tok/s	Q4_K_M · 13.1 GB
4	Qwen3.6 35B A3BAlibaba Q4_K_M · 22.0 GB · 17.7 t/s	17.7 tok/s	Q4_K_M · 22.0 GB
5	LFM2.5 24B A2BLiquid AI Q4_K_M · 15.0 GB · 29.9 t/s	29.9 tok/s	Q4_K_M · 15.0 GB

#	Model	Speed ↓	Quant · Memory
1	Qwen3 30B A3B Thinking 2507AlibabaHow to run → Q4_K_M · 19.4 GB · 22.7 t/s	22.7 tok/s	Q4_K_M · 19.4 GB
2	NVIDIA Nemotron 3.5 Lightning 30B A3B BF16NVIDIA Q4_K_M · 19.3 GB · 28.9 t/s	28.9 tok/s	Q4_K_M · 19.3 GB
3	Qwen3 30B A3B Instruct 2507Alibaba Q4_K_M · 19.4 GB · 22.7 t/s	22.7 tok/s	Q4_K_M · 19.4 GB
4	GPT OSS 20BOpenAI Q4_K_M · 13.1 GB · 20.8 t/s	20.8 tok/s	Q4_K_M · 13.1 GB
5	Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 19.2 t/s	19.2 tok/s	Q4_K_M · 16.5 GB

#	Model	Speed	Quant · Memory
1	Qwen3.6 35B A3BAlibabaHow to run → Q4_K_M · 22.0 GB · 17.7 t/s	17.7 tok/s	Q4_K_M · 22.0 GB
2	Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 19.2 t/s	19.2 tok/s	Q4_K_M · 16.5 GB
3	Qwen3.5 4BAlibaba Q4_K_M · 3.4 GB · 18.3 t/s	18.3 tok/s	Q4_K_M · 3.4 GB
4	Gemma 3 4B ITGoogle Q4_K_M · 5.0 GB · 21.0 t/s	21.0 tok/s	Q4_K_M · 5.0 GB
5	MiniCPM V 4.7 35B A3BOpenBMB Q4_K_M · 22.1 GB · 20.4 t/s	20.4 tok/s	Q4_K_M · 22.1 GB

See all 329 models on the Intel Core Ultra 9 285H (Arrow Lake) Laptop →Bars = traction for models without a Score yet

About the Intel Core Ultra 9 285H (Arrow Lake) Laptop

The Intel Core Ultra 9 285H (Arrow Lake) Laptop is a 32 GB laptop with more memory than speed: it runs 35B MoE models well. The system keeps about 8 GB, leaving 24 GB for models. With 39 TFLOPS it reads long prompts faster than most 32 GB machines: a 4K-token prompt takes 5.6 s. Core Ultra 9 285H, released 2025.

Speed in 32 GB#12120 GB/s

Prompt reading#339 TFLOPS

For models24 GBThe system keeps 8 GB

ChipCore Ultra 9 285Hreleased 2025

Intel Core Ultra 9 285H (Arrow Lake) Laptop strengths and weaknesses

Strengths

Runs 35B MoE models: Qwen3.6 35B A3B at 17.7 tok/s in 22 GB
Low power draw: 45 W at full load for the whole machine

Weaknesses

70B models don't fit: Llama 3.3 70B Instruct needs 51.3 GB at Q4 vs 24 GB usable here
Slow on large dense models: Qwen3.8 27B at 3.4 tok/s
Slow prompt reading: A 32K-token prompt on Qwen3.6 35B A3B waits 45.0 s for the first token
oneAPI, not CUDA: No FP8 or NVFP4 weights and no vLLM AWQ or GPTQ; use GGUF builds

Memory for models

24 GB of 32 GB

9B6.6 GB14B11.9 GB27–31B17.9 GB35B MoE22 GB

9B6.6 GB14B11.9 GB35B MoE22 GB

For models 24 GBThe system keeps 8 GBModel size at Q4

70B models need 51.3 GB at Q4, more than the 24 GB available.Or rent an A100 (80 GB) from $1.06/hr →

Similar devices and alternatives

The same chip from other makers first, then machines and GPUs people weigh against it, each with the one reason it is worth a look. Speeds on Qwen3.6 35B A3B.

162% faster, 3.5 GB less memoryMac Mini M4 Pro (24 GB)Apple · 20.5 GB for models · M4 Pro

▲ 162% speed▼ 3.5 GB$1,599

45% faster, 3.5 GB less memoryMacBook Air 13" M4 (24 GB)Apple · 20.5 GB for models · M4

▲ 45% speed▼ 3.5 GB$1,399

75% faster, 4.5 GB more memoryMac Mini M6 (32 GB)Apple · 28.5 GB for models · M6

▲ 75% speed▲ 4.5 GB$1,299

62% faster, 4.5 GB more memoryMacBook Pro 14 M5Apple · 28.5 GB for models · M5

▲ 62% speed▲ 4.5 GB$1,839

Reads prompts 11× faster, 24 GB for modelsRTX 4090NVIDIA · 24 GB VRAM · GPU

▲ 11× prompts▲ 753% speedsame memory$1,599rent $0.34/hr

Reads prompts 14× faster, 32 GB for modelsRTX 5090NVIDIA · 32 GB VRAM · GPU

▲ 14× prompts▲ 1105% speed▲ 8 GB$5,502rent $0.50/hr

Intel Core Ultra 9 285H (Arrow Lake) Laptop tokens per second by model size

Output speed depends on memory bandwidth (120 GB/s). Grey rows need more than the 24 GB available.

9BQwen3.5 9B · Q4_K_M · 6.6 GB slow
9.4tok/s
14BQwen3 14B · Q4_K_M · 11.9 GB slow
6.3tok/s
27–31BQwen3.8 27B · Q4_K_M · 17.9 GB slow
3.4tok/s
35B MoEQwen3.6 35B A3B · Q4_K_M · 22.0 GB
17.7tok/s
70BLlama 3.3 70B Instruct · not at 4-bit
needs 51.3 GB
120B MoEGPT OSS 120B · not at 4-bit
needs 70.8 GB

Estimated speeds: Q4_K_M, 4K context, one user. How we estimate

Unified-memory platforms: writing vs compute

Writing, tok/sCompute, TFLOPS

M116 GB · 68.3 GB/s
24.9
5
Ryzen 7 8845HS32 GB · 89.6 GB/s
16.2
8
M216 GB · 100 GB/s
33.4
7
Orin NX16 GB · 102.4 GB/s
37.0
15
M316 GB · 102.4 GB/s
34.0
7
Core Ultra 9 285H32 GB · 119.5 GB/s
17.7
39

Machines with about the same memory bandwidth write at about the same speed; compute decides how fast prompts are read. Speeds on Qwen3.6 35B A3B.

All 1158 models on the Intel Core Ultra 9 285H (Arrow Lake) Laptop

Featured only

Best quant that fits in 24 GB, ordered by llmrun Score. Fit grades how well it runs here, S best. How ranking works

LLM models compatible with Intel Core Ultra 9 285H (Arrow Lake) Laptop — ranked by Score, recency and Fit
Model	Quant	VRAM	Speed	Context	Fit
GPT OSS Safeguard 20B21.5BChatFunctions Q4_K_M·18.4 t/s	Q4_K_M	13.5 GB56%	18.4 t/s	131K	B58, Decent
Yi Coder 9B Chat8.8BChatCode Q4_K_M·10.3 t/s	Q4_K_M	7.2 GB30%	10.3 t/s	131K	B49, Decent
Luciole 1B Base1.3BChat Q4_K_M·50.2 t/s	Q4_K_M	1.9 GB8%	50.2 t/s	131K	A79, Runs well
Kolibri 1 BF1678.1BChatReasoningFunctions IQ2_XXS·38.2 t/s	IQ2_XXS	22.5 GB94%	38.2 t/s	262K	B53, Decent
Llama 3.1 8B8.0BChat Q4_K_M·11.3 t/s	Q4_K_M	7.9 GB33%	11.3 t/s	—	B50, Decent
Llama 3.1 8B Lexi Uncensored v28.0BChat Q4_K_M·11.1 t/s	Q4_K_M	7.3 GB30%	11.1 t/s	131K	B49, Decent
Nemotron Labs Diffusion 14B13.5BChat Q4_K_M·6.8 t/s	Q4_K_M	11.8 GB49%	6.8 t/s	262K	C44, Usable
Qwen2.5 7B Instruct7.6BChatFunctions Q4_K_M·12.0 t/s	Q4_K_M	5.8 GB24%	12.0 t/s	33K	B51, Decent
Gemma 3 270M IT270MChat Q4_K_M·331.9 t/s	Q4_K_M	1.2 GB5%	331.9 t/s	—	S100, Runs great
Qwen2.5 Math 1.5B Instruct1.5BChatMathFunctions Q4_K_M·46.7 t/s	Q4_K_M	1.3 GB6%	46.7 t/s	4K	A78, Runs well
Qwen3 Code Reasoning 4B4BChatCodeReasoning Q4_K_M·20.7 t/s	Q4_K_M	4.2 GB18%	20.7 t/s	262K	B61, Decent
Aya Expanse 8B8.0BChat BF16·3.4 t/s	BF16	20.2 GB84%	3.4 t/s	—	C34, Usable
Qwen3 8B Base8.2BChat Q4_K_M·10.8 t/s	Q4_K_M	7.6 GB32%	10.8 t/s	33K	B49, Decent
MiniCPM 2B Sft BF162BChat Q4_K_M·26.6 t/s	Q4_K_M	3 GB13%	26.6 t/s	4K	A67, Runs well
Goedel Prover v2 32B32.8BChat Q4_K_M·2.9 t/s	Q4_K_M	22.6 GB94%	2.9 t/s	41K	D18, Barely runs

Intel Core Ultra 9 285H (Arrow Lake) Laptop specs and power

Specs

Chip: Core Ultra 9 285H · Intel Arc 140T
CPU cores: 16
GPU cores: 8
Unified memory: 32 GB LPDDR5X-7467 (soldered, 128-bit) in most thin-and-light builds
For models: 24 GB (The system keeps 8 GB)
Memory bandwidth: 119.5 GB/s
FP16 (dense): 38.5 TFLOPS
NPU: Intel AI Boost NPU 3 (13 TOPS)
Power: 45 W max
Released: 2025-01-06

Intel Core Ultra 9 285H (Arrow Lake) Laptop FAQ

How much memory is available for AI on the Intel Core Ultra 9 285H (Arrow Lake) Laptop?

About 24 GB of the 32 GB. The system keeps roughly 8 GB for itself and other apps.

What is the largest model the Intel Core Ultra 9 285H (Arrow Lake) Laptop can run?

Qwen3.6 35B A3B (36.0B) is the largest at 4-bit or better: at Q4_K_M it uses 22.0 GB, 92% of the available memory, and runs at about 17.7 tok/s.

Can the Intel Core Ultra 9 285H (Arrow Lake) Laptop run 70B models?

No. Llama 3.3 70B Instruct needs 51.3 GB at Q4, more than the 24 GB available.

How long does the Intel Core Ultra 9 285H (Arrow Lake) Laptop take to read a long prompt?

On Qwen3.6 35B A3B, about 5.6 s for a 4K-token prompt and 45 s for a 32K-token document before the first word appears.

Is renting a GPU cheaper than buying the Intel Core Ultra 9 285H (Arrow Lake) Laptop?

An RTX 5090 (32 GB) rents from $0.46/hr.

24 GB GPUs Laptops Mac Studio M1 Max Mac Studio M2 Max MacBook Pro 14" M1 Max MacBook Pro 14" M2 Max MacBook Pro 16" M1 Max MacBook Pro 16" M2 Max Intel Core Ultra X9 388H (Panther Lake) Laptop Best coding LLMs All hardware