Downloads · 30 days
39
21% of all-time downloads
Qwen3.5-35B-A3B-oQ2-mtp
mlx-works/Qwen3.5-35B-A3B-oQ2-mtp
Qwen3.5-35B-A3B-oQ2-mtp is a machine learning model from mlx-works. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for mlx.
This model was quantized using oQ (oMLX v0.5.2.dev1) mixed-precision quantization.
All-time downloads
186
Public
Parameters
36B
13.6 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors13.6 GB · 100%
Parameter types
How the weights are stored.
U3235.5B · 99%
Model card
From the Hugging Face model README
Qwen3.5-35B-A3B-oQ2-mtp
This model was quantized using oQ (oMLX v0.5.2.dev1) mixed-precision quantization.
Base model: Qwen/Qwen3.5-35B-A3B
Chat template: froggeric/Qwen-Fixed-Chat-Templates
Quantization details
- Model type: qwen3_5_moe
- Bits: 2
- Group size: 64
- Format: MLX safetensors
Environment
- Hardware: M5 MacBook Air 32GB
- Inference Framework: oMLX v0.5.2.dev1
- Max Concurrent Requests: 4
- Settings:
- Thinking: Disabled
- Chat template parameter: enable_thinking=false (forced)
- TurboQuant KV Cache: Enabled (disable before oMLX v0.5.2.dev1)
- Lightning MTP: Enabled (key speed improvement)
Performance Benchmarks
Note: Results are for reference only and may vary depending on hardware, software configuration, and workload.
Single Request Results
| Test | TTFT(ms) | TPOT(ms) | pp TPS | tg TPS | E2E(s) | Throughput | Peak Mem |
|---|---|---|---|---|---|---|---|
| pp1024/tg128 | 1052.6 | 18.34 | 972.8 tok/s | 55.0 tok/s | 3.398 | 339.1 tok/s | 13.04 GB |
| pp4096/tg128 | 3671.7 | 18.96 | 1115.6 tok/s | 53.2 tok/s | 6.095 | 693.0 tok/s | 13.75 GB |
Continuous Batching (pp1024 / tg128)
| Batch | tg TPS | Speedup | pp TPS | pp TPS/req | TTFT(ms) | E2E(s) |
|---|---|---|---|---|---|---|
| 1x | 55.0 tok/s | 1.00x | 972.8 tok/s | 972.8 tok/s | 1052.6 | 3.398 |
| 2x | 79.4 tok/s | 1.44x | 884.7 tok/s | 442.4 tok/s | 2314.2 | 5.540 |
| 4x | 111.2 tok/s | 2.02x | 870.2 tok/s | 217.6 tok/s | 4540.7 | 9.310 |
Intelligence Benchmark
Note: Each benchmark round tests only 30 questions. Results are for reference only.
| Benchmark | Accuracy | Correct | Total | Time(s) | Think |
|---|---|---|---|---|---|
| MMLU | 70.0% | 21 | 30 | 30.5 | No |
| TRUTHFULQA | 83.3% | 25 | 30 | 14.8 | No |
| GSM8K | 93.3% | 28 | 30 | 104.9 | No |
| MATHQA | 53.3% | 16 | 30 | 20.4 | No |
| HUMANEVAL | 83.3% | 25 | 30 | 124.5 | No |
- Library
- mlx
- Type
- qwen3_5_moe
- Created
- Jul 18, 2026
- Updated
- Jul 19, 2026