← Leaderboard
Qwen3.6-35B-A3B Unsloth Dynamic UD-Q8_K_XL MLX (Brooooooklyn)
Keeper Rank #8 of 33 · 5/8 GAUNTLET progress
Hollow markers & dashed spokes: axis not yet scored
G
93.5
A
95
U
—
N
91.2
T
100
L
—
E
—
T
41.9
Specification
- Parameters
- 35B total / 3B active per token
- Architecture
- qwen3_5_moe
- Size on disk
- 38.06 GB
- Quantization
- UD-Q8_K_XL mixed-precision
- Format
- MLX
- Reasoning (CoT)
- Yes — emits reasoning tokens
- Internal ID
- M9b
- Mean speed
- 68.4 tok/s across suites
- Stall census
- 0 stalls in 30 observed tests (0.0%)
- Reasoning appetite
- 2,240 tokens mean · 6,384 max
- Model card
- huggingface.co
Suite results
| Suite | Score | Avg / 20 | Tests | tok/s |
|---|---|---|---|---|
| General capability (13-task real-workload suite) 1a | 243.1 / 260 | 18.7 | 13/13 | 83.1 |
| Agentic tool-calling & protocol adherence 1b | 152 / 160 | 19 | 8/8 | 81.3 |
| Long-context retrieval & synthesis 1g | 120 / 120 | 20 | 6/6 | 59.8 |
| Long-context multi-needle (MRCR) 1g2 | 44.1 / 60 | 14.7 | 3/3 | 49.5 |
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.