What hardware do I need?

Hardware that delivers the experience you want — not just loads the model

Pick the model and describe the workload. We rate every machine in the database for that exact workload and group them by whether they meet your target.

1 · Model

Quantization

2 · Use case

Agentic coding

Chat & writing

Coding

Documents & research

Multimodal

Automation & serving

3 · Tool & runtime

Tool: OpenCode

Runtime

4 · Workload

Context

Project size, agent behaviour, other apps

Coding style

Repository size

Repository size is a signal for retrieval and prompt-processing load — the whole repo never goes into the context.

How does the agent behave?

What else do you normally run while coding?

This memory is subtracted before the model is considered. Very important on 16–32 GB machines.

5 · Desired experience

6 · Budget & vendors (optional)

Prices are approximate launch prices for the whole system.

USD max

AppleNVIDIAAMDIntel

Qwen3.5 35B-A3B · Agentic coding / medium repository

Target: comfortable · 54 machines meet it

Meets your comfortable target (54)

Cheapest first.

Comfortable

Ryzen AI Max+ 395 mini PC / Framework Desktop · 64 GB

64 GB unified memory · 256 GB/s

Quant / runtime: Q6 · Lemonade
Generation: 50–65 tok/s
Per step: ≈15 s
Headroom: 20 GB

$1,600Calibrated

Comfortable

Mac mini M4 Pro 48 GB

48 GB unified memory · 273 GB/s

Quant / runtime: Q5 · LM Studio
Generation: 50–70 tok/s
Per step: ≈15 s
Headroom: 12 GB

$1,799Calibrated

Comfortable

Ryzen AI Max+ 395 mini PC / Framework Desktop · 128 GB

128 GB unified memory · 256 GB/s

Quant / runtime: Q6 · Lemonade
Generation: 50–65 tok/s
Per step: ≈15 s
Headroom: 84 GB

$2,000Calibrated

Comfortable

Mac mini M5 Pro 48 GB

48 GB unified memory · 307 GB/s

Quant / runtime: Q5 · LM Studio
Generation: 45–75 tok/s
Per step: ≈10 s
Headroom: 12 GB

$2,099Estimated

Comfortable

Mac mini M4 Pro 64 GB

64 GB unified memory · 273 GB/s

Quant / runtime: Q6 · LM Studio
Generation: 45–60 tok/s
Per step: ≈15 s
Headroom: 22 GB

$2,199Calibrated

Comfortable

Desktop PC · Radeon AI PRO R9700 · 32 GB RAM

32 GB VRAM (640 GB/s) + 32 GB RAM

Quant / runtime: Q6 · Lemonade
Generation: 120–170 tok/s
Per step: ≈4 s
Headroom: 1.6 GB

$2,200Calibrated

Comfortable

Desktop PC · Radeon AI PRO R9700 · 64 GB RAM

32 GB VRAM (640 GB/s) + 64 GB RAM

Quant / runtime: Q6 · Lemonade
Generation: 120–170 tok/s
Per step: ≈4 s
Headroom: 1.6 GB

$2,328Calibrated

Comfortable

Mac Studio M1 Max 64 GB

64 GB unified memory · 400 GB/s

Quant / runtime: Q6 · LM Studio
Generation: 60–80 tok/s
Per step: ≈10 s
Headroom: 22 GB

$2,399Calibrated

Comfortable

Mac mini M5 Pro 64 GB

64 GB unified memory · 307 GB/s

Quant / runtime: Q6 · LM Studio
Generation: 40–65 tok/s
Per step: ≈10 s
Headroom: 22 GB

$2,499Estimated

Meets the acceptable target (5)

Usable, with noticeable compromises for this workload.

Acceptable

MacBook Pro M3 Pro 36 GB

36 GB unified memory · 150 GB/s

Quant / runtime: Q4 · LM Studio
Generation: 35–45 tok/s
Per step: ≈20 s
Headroom: 5.1 GB

$2,399Calibrated

Acceptable

MacBook Pro M3 Max 36 GB

36 GB unified memory · 300 GB/s

Quant / runtime: Q4 · LM Studio
Generation: 65–90 tok/s
Per step: ≈10 s
Headroom: 5.1 GB

$3,199Calibrated

Acceptable

MacBook Pro M4 Max 36 GB

36 GB unified memory · 410 GB/s

Quant / runtime: Q4 · LM Studio
Generation: 80–110 tok/s
Per step: ≈8 s
Headroom: 5.1 GB

$3,199Calibrated

Acceptable

MacBook Pro M5 Max 36 GB

36 GB unified memory · 460 GB/s

Quant / runtime: Q4 · LM Studio
Generation: 80–130 tok/s
Per step: ≈6 s
Headroom: 5.1 GB

$3,599Estimated

Acceptable

Desktop PC · Ryzen 9 9950X (no GPU) · 64 GB DDR5

64 GB RAM · 89 GB/s · CPU only

Quant / runtime: Q4 · llama.cpp (llama-server)
Generation: 12–20 tok/s
Per step: ≈40 s
Headroom: 28 GB

$1,400Estimated

Can run, but below your target (37)

The model loads, but the experience falls short for this workload.

Cannot run (32)

Not enough usable memory, or no compatible runtime.