Hollow markers & dashed spokes: axis not yet scored
G
92.5
A
95
U
—
N
90.2
T
100
L
—
E
—
T
24.1
Specification
- Parameters
- 122B total / 10B active per token
- Architecture
- qwen3_5_moe
- Size on disk
- 69.62 GB
- Quantization
- 4bit
- Format
- MLX
- Reasoning (CoT)
- Yes — emits reasoning tokens
- Internal ID
- M10
- Mean speed
- 39.4 tok/s across suites
- Stall census
- 0 stalls in 30 observed tests (0.0%)
- Reasoning appetite
- 2,043 tokens mean · 9,288 max
- Model card
- huggingface.co
Suite results
| Suite | Score | Avg / 20 | Tests | tok/s |
|---|---|---|---|---|
| General capability (13-task real-workload suite) 1a | 240.5 / 260 | 18.5 | 13/13 | 53.6 |
| Agentic tool-calling & protocol adherence 1b | 152 / 160 | 19 | 8/8 | 43.8 |
| Long-context retrieval & synthesis 1g | 118.2 / 120 | 19.7 | 6/6 | 35.9 |
| Long-context multi-needle (MRCR) 1g2 | 44.1 / 60 | 14.7 | 3/3 | 24.3 |
Each test is LLM-judged 0–20 against a fixed rubric; suite max = tests × 20. Where fewer tests ran than the suite total, unrun tests count as zero toward the axis score. See methodology.