← Leaderboard
Qwen3.8-27B MTPLX Optimized-Speed 4bit MLX (Youssofal)
Retired Rank #49 of 53 · 2/8 GAUNTLET progress
Hollow markers & dashed spokes: axis not yet scored
G
68.1
A
—
U
—
N
—
T
n/d
L
—
E
—
T
6.5
Specification
- Parameters
- 27B dense
- Architecture
- qwen3_5
- Size on disk
- 20.4 GB
- Quantization
- 4bit
- Format
- MLX
- Runtime
- LM Studio
- Class
- Standard
- Reasoning (CoT)
- Yes — emits reasoning tokens
- Internal ID
- M25e
- Mean speed
- 22.2 tok/s across suites
- Stall census
- 1 stall in 1 observed tests (100.0%)
- Reasoning appetite
- 10,476 tokens mean · 10,476 max
- Model card
- huggingface.co
Suite results
Suite Score Avg / 20 Tests n tok/s Run
General capability (13-task real-workload suite) 1a 177 / 260 17.7 10/13 ● n=1 22.2 run 2026-09-12
| Test | Category | Score | tok/s | |
|---|---|---|---|---|
| A2 | Dev/Ops Scripting | 18/20 | 23.9 | |
| A3 | Dev/Ops Scripting | 20/20 | 22.7 | |
| A4 | Dev/Ops Scripting | 19/20 | 21.3 | |
| A5 | Dev/Ops Scripting | 20/20 | 22.8 | |
| A6 | Dev/Ops Scripting | 20/20 | 23.3 | |
| B1 | Document Processing | 20/20 | 22.8 | |
| B2 | Document Processing | 20/20 | 22.7 | |
| C2 | Content Production | 20/20 | 20.9 | |
| C4 | Content Production | 20/20 | 21.3 |
Category mean Dev/Ops Scripting 19.4Document Processing 20Content Production 20
Each test is LLM-judged 0–20 against a fixed rubric; suite max = tests × 20. An ADJ chip marks a cell carrying a subjective adjustment from the maintainer; the judged score is the one shown, is unchanged, and remains what the axes, ranks and tie bands are computed from. Hover for the reason. Rows with a ▶ expand to the per-test breakdown. Where fewer tests ran than the suite total, unrun tests count as zero toward the axis score. See methodology.