Hollow markers & dashed spokes: axis not yet scored
G
78
A
97
U
—
N
—
T
100
L
—
E
—
T
6.3
Specification
- Parameters
- 24B
- Architecture
- mistral3
- Size on disk
- 19.43 GB
- Quantization
- 6bit
- Format
- MLX
- Runtime
- LM Studio
- Class
- Standard
- Reasoning (CoT)
- No
- Internal ID
- M22
- Mean speed
- 22 tok/s across suites
- Stall census
- 0 stalls in 22 observed tests (0.0%)
- Model card
- huggingface.co
Suite results
Suite Score Avg / 20 Tests n tok/s Run
General capability (13-task real-workload suite) 1a 202.8 / 260 15.6 13/13 n=1.1 17.7 run 2026-08-06
| Test | Category | Score | tok/s | |
|---|---|---|---|---|
| A1 | Dev/Ops Scripting | 14/20 | 26.1 | |
| A2 | Dev/Ops Scripting | 14/20 | 24.9 | |
| A3 | Dev/Ops Scripting | 9/20 | 23.5 | |
| A4 | Dev/Ops Scripting | 18/20 | 23.8 | |
| A5 | Dev/Ops Scripting | 16/20 | 18.4 | |
| A6 | Dev/Ops Scripting | 20/20 | 16.8 | |
| B1 | Document Processing | 20/20 | 14.2 | |
| B2 | Document Processing | 20/20 | 15 | |
| B3 | Document Processing | 12/20 | 13.8 | |
| C1 | Content Production | 15/20 | 13.3 | |
| C2 | Content Production | 15/20 | 12.9 | |
| C3 | Content Production | 18/20 | 13.7 | |
| C4 | Content Production | 12/20 | 14.1 |
Category mean Dev/Ops Scripting 15.2Document Processing 17.3Content Production 15
Agentic tool-calling & protocol adherence 1b 155.2 / 160 19.4 8/8 n=1 26.3 run 2026-08-06
| Test | Category | Score | tok/s | |
|---|---|---|---|---|
| WA1 | Tool Calling & Protocol Adherence | 20/20 | 20.1 | |
| WA2 | Tool Calling & Protocol Adherence | 20/20 | 22 | |
| WA3 | Tool Calling & Protocol Adherence | 20/20 | 22 | |
| WA4 | Tool Calling & Protocol Adherence | 20/20 | 20.7 | |
| WA5 | Tool Calling & Protocol Adherence | 20/20 | 25.3 | |
| WB1 | Agent Robustness & State Management | 20/20 | 25 | |
| WB2 | Agent Robustness & State Management | 20/20 | 24.6 | |
| WB3 | Agent Robustness & State Management | 15/20 | 23.8 |
Category mean Tool Calling & Protocol Adherence 20Agent Robustness & State Management 18.3
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.