Qwythos-9B-v2-MLX-oQ8-mtp
This model was quantized using oQ (oMLX v0.5.4.dev1) mixed-precision quantization.
Quantization details
- Model type: qwen3_5
- Bits: 8
- Group size: 64
- Format: MLX safetensors
Testing
I used these params recently for a complex reasoning task requring RAG and high-level thinking. Results were slow but exceptionally strong.
Added kwargs forced reasoning effort = Max
ACTIVE MODEL
Qwythos-9B-v2-MLX-oQ8-mtp
TEMPERATURE
0.6
MAX TOKENS
16384
MIN P
0.95 TOP K 20
REP. PENALTY
1.05
PRESENCE PENALTY
Default
THINKING
On (Unlimited)
- Downloads last month
- 309
Model size
3B params
Tensor type
BF16
·
U32 ·
Hardware compatibility
Log In to add your hardware
8-bit
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for brainworkup/Qwythos-9B-v2-MLX-oQ8-mtp
Base model
Qwen/Qwen3.5-9B-Base Finetuned
Qwen/Qwen3.5-9B Finetuned
empero-ai/Qwythos-9B-Claude-Mythos-5-1M Finetuned
empero-ai/Qwythos-9B-v2