Qwen3.5 2B: RAM, speed and settings on a Mac
Qwen3.5 2B is the smallest chat model in PocketWebTools for Mac: a 1.3 GB file that needs about 2.0 GB of memory at the app's 8K context, so it fits any Mac the app supports. Below is what we measured running it on a 16 GB M1 Pro, and the settings the app uses.
Which Mac runs it
The chat models the app ships, by the memory your Mac has. Loaded is the file plus the buffers llama.cpp allocates at the app's context window. Speeds are from a 16 GB M1 Pro; the two 32 GB models ran on a 32 GB Mac mini M4. Speed depends mostly on the chip, not the memory size, so another Mac can be faster or slower than ours. The app shows the speed of every reply on your own Mac.
| Model | Mac memory | Download | Loaded | Reply speed |
|---|---|---|---|---|
| Qwen3.8 27B | 32 GB and up | 17.6 GB | 20.4 GB | 5.3 to 5.8 tokens/sMac mini M4 |
| Qwen3.6 35B‑A3B | 32 GB and up | 22.4 GB | 23.7 GB | 25.1 to 27.3 tokens/sMac mini M4 |
| Qwen3.5 9B | 16 GB and up | 6.0 GB | 7.1 GB | 18.6 to 19 tokens/sM1 Pro |
| Gemma 4 12B | 16 GB and up | 7.4 GB | 8.8 GB | 13.8 to 14.7 tokens/sM1 Pro |
| Bonsai 27B (1‑bit) | 8 GB and up | 3.8 GB | 5.1 GB | 14 to 16.3 tokens/sM1 Pro |
| Qwen3.5 2B | 8 GB and up | 1.3 GB | 2.0 GB | 73 to 76 tokens/sM1 Pro |
Qwen3.5 2B speed on an M1 Pro
Measured in the app's own engine: llama.cpp on Metal, every layer on the GPU, other apps closed. Speed depends mostly on the chip, not the memory size, so another Mac can be faster or slower than ours. The app shows the speed of every reply on your own Mac.
- Load, file already on disk
- 0.4 s
The first load after a restart, reading the file from disk, took 22 s on our Mac.
- Reply speed
- 73 to 76 tokens/s
Qwen3.5 9B replied at 18.6 to 19.0 tokens per second on the same Mac.
- Reading a prompt
- About 1,000 tokens/s
A 774-token prompt took 0.76 s before the first word appeared.
- Follow-up in the same chat
- 0.05 s to the first word
440 of 471 prompt tokens came from the cache.
- Our nine-prompt test
- 9 of 9
Summarizing, rewriting, questions about an 800-token memo, list formatting, TypeScript and a reasoning puzzle. It scored 9 of 9 again on 24 October.
Settings the app runs Qwen3.5 2B with
The same values work in any llama.cpp front end.
- File
- Qwen3.5-2B-UD-Q4_K_XL.gguf
Unsloth's 4‑bit dynamic quantization, pinned to one revision and checked against its SHA-256 after download.
- Context window
- 8,192 tokens
- GPU layers
- All, on Metal
- Batch size
- 512
- Temperature
- 0.7
- Top-p / top-k / min-p
- 0.9 / 40 / 0.05
- Thinking
- Off
The reply starts after an empty think block, so the answer begins straight away.
- Prompt cache
- Kept between turns
A follow-up only reads the new message, not the whole chat again.
What the app does with Qwen3.5 2B
A first model while a bigger one downloads
It downloads in a couple of minutes, so you can start chatting while a bigger model is still on its way.
Chat
Threads saved on your Mac, replies rendered as Markdown with code blocks and tables, and a meter that shows how much of the 8K window a thread has used.
Clip
On an 8 GB Mac the app picks clips with it first: about 5 seconds for a window of transcript, where Bonsai 27B took 45.
Where it falls short
- Every speed here comes from one 16 GB M1 Pro.
- It picks clips fast but not well: on our test podcast it found 1 of 4 planted highlights and started that one 10 seconds late. Bonsai 27B found 2 of 4 on the same video.
- The app runs it at an 8K context, so a long document may not fit in one chat.
- In the app it works on text only.
- Nine prompts is a small test. It checks everyday writing and reading jobs; it is not a leaderboard.
Run Qwen3.5 2B in PocketWebTools for Mac
One app for chat, transcription, documents, voices, photos and video, with every model on your Mac. $99 once, free lifetime updates.
Already have a key? Download the app. It asks for the key the first time it opens.
14-day money-back guarantee. Your license key is shown right after checkout and stays on your purchase page; one key works on up to 3 Macs.
Qwen3.5 2B questions
- How much RAM does Qwen3.5 2B need?
- About 2.0 GB at an 8,192-token context: the 1.3 GB file (Unsloth's UD-Q4_K_XL quantization) plus 0.65 GB of buffers, measured from llama.cpp's own log on a 16 GB M1 Pro. PocketWebTools for Mac files it under 8 GB Macs.
- How fast is Qwen3.5 2B on an M1 Pro?
- 73 to 76 tokens per second when replying and about 1,000 tokens per second when reading a prompt, on a 16 GB M1 Pro with every layer on the GPU. A follow-up in the same chat starts in about 0.05 seconds. Another Mac can be faster or slower: speed depends mostly on the chip, and the app shows your own speed under every reply.
- Qwen3.5 2B or Bonsai 27B on an 8 GB Mac?
- Qwen3.5 2B for speed and room to spare: about 2.0 GB loaded and 73 to 76 tokens per second on our M1 Pro, against 5.1 GB and 14 to 16 for Bonsai 27B. Both did well on our nine prompts (9 of 9 and 8.5). Bonsai found more of the planted highlights in our clip test, at about nine times the wait.
- Can I use Qwen3.5 2B commercially?
- Yes. It is released under Apache 2.0, which allows commercial use. The app ships the license text and adds no restrictions of its own.