REAL LOCAL SUITES · CODING V2 + COMPREHENSIVE · SUITES NOT MIXED
RUN REGISTRY
Every run keeps its classification
Real local suite captures, performance probes, and fictional mock fixtures remain separate records.
LOCAL SUITE RUNS
Coding v2 + Comprehensive
| Run ID | Suite | Model | Score | Gen tok/s | Date | Status |
|---|---|---|---|---|---|---|
HB-20260912-C9B4A1 LOCAL SUITE | Coding v2 | official Qwen3.5-9B Q4_K_MQ4_K_M | 9/12 | 13.92 | 2026-09-12 | Installed |
HB-20260912-N4B001 LOCAL SUITE | Coding v2 | Nemotron3 Nano 4B Q5_K_MQ5_K_M | 8/12 | 23.52 | 2026-09-12 | Installed |
HB-20260912-G4F8A2 LOCAL SUITE | Coding v2 | gemma4-fable5Q5_K_M | 8/12 | 10.81 | 2026-09-12 | Installed |
HB-20260912-HQ8B01 LOCAL SUITE | Coding v2 | huihui Qwen3 abliterated 8BQ4_K_M | 8/12 | 17.27 | 2026-09-12 | Installed |
HB-20260912-Q34B01 LOCAL SUITE | Coding v2 | Qwen3-4B-2507 HauhauCS Q6_KQ6_K | 7/12 | 23.79 | 2026-09-12 | Removed |
HB-20260912-MN3B01 LOCAL SUITE | Coding v2 | Ministral-3-3B-Instruct-2512 Q5_K_MQ5_K_M | 6/12 | 30.91 | 2026-09-12 | Removed |
HB-20260912-HQ4B01 LOCAL SUITE | Coding v2 | HauhauCS Qwen3.5-4BQ4_K_M | 5/12 | 26.86 | 2026-09-12 | Installed |
HB-20260912-QW9B01 LOCAL SUITE | Coding v2 | qwythos-9b-nsfwQ4_K_M | 4/12 | 16.78 | 2026-09-12 | Installed |
HB-20260912-RO1201 LOCAL SUITE | Coding v2 | rocinante-nsfwQ4_K_M | 4/12 | 13.21 | 2026-09-12 | Installed |
HB-20260911-HQ4BC80 LOCAL SUITE | Comprehensive | HauhauCS Qwen3.5-4BQ4_K_M | 49/80 | 26.09 | 2026-09-11 | Installed |
HB-20260911-Q3827B LOCAL SUITE | Comprehensive | Qwen3.8-27B UD-IQ2_XXSUD-IQ2_XXS | 45.5/80 | 6.64 | 2026-09-11 | Removed |
HB-20260911-QW9B80 LOCAL SUITE | Comprehensive | qwythos-9b-nsfwQ4_K_M | 42.5/80 | 16.82 | 2026-09-11 | Installed |
HB-20260912-G4F80 LOCAL SUITE | Comprehensive | gemma4-fable5Q5_K_M | 41.5/80 | 6.87 | 2026-09-12 | Installed |
HB-20260911-O9B80 LOCAL SUITE | Comprehensive | official Qwen3.5-9B Q4_K_MQ4_K_M | 41/80 | 9.28 | 2026-09-11 | Installed |
HB-20260912-HQ8B80 LOCAL SUITE | Comprehensive | huihui Qwen3 abliterated 8BQ4_K_M | 40/80 | 19.88 | 2026-09-12 | Installed |
HB-20260912-N4B80 LOCAL SUITE | Comprehensive | Nemotron3 Nano 4B Q5_K_MQ5_K_M | 39.5/80 | 22.28 | 2026-09-12 | Installed |
HB-20260912-Q34B80 LOCAL SUITE | Comprehensive | Qwen3-4B-2507 HauhauCS Q6_KQ6_K | 39/80 | 23.60 | 2026-09-12 | Removed |
HB-20260911-HQ9B80 LOCAL SUITE | Comprehensive | HauhauCS Qwen3.5-9B Q6_KQ6_K | 38.5/80 | 7.89 | 2026-09-11 | Removed |
HB-20260912-MN3B80 LOCAL SUITE | Comprehensive | Ministral-3-3B-Instruct-2512 Q5_K_MQ5_K_M | 38.5/80 | 30.91 | 2026-09-12 | Removed |
HB-20260911-UH9B80 LOCAL SUITE | Comprehensive | ultra-heretic 9B Q6_KQ6_K | 35.5/80 | 7.86 | 2026-09-11 | Removed |
HB-20260912-G4E80 LOCAL SUITE | Comprehensive | HauhauCS Gemma-4-E4B IQ4_XSIQ4_XS | 34.5/80 | 29.95 | 2026-09-12 | Removed |
HB-20260912-RO1280 LOCAL SUITE | Comprehensive | rocinante-nsfwQ4_K_M | 26.5/80 | 10.45 | 2026-09-12 | Installed |
REAL PERFORMANCE RUNS
Reliable performance methodology
| Run ID | Model | Status | Warm median | Samples | Outliers | Cleanup | Leaderboard |
|---|---|---|---|---|---|---|---|
HB-20260908-1BE78F REAL PERFORMANCE RUN | moondream:latestQ4_0 | PARTIAL | NOT COLLECTED | 0/5 | 0 | verified | INELIGIBLE |
HB-20260908-1A4AF9 REAL PERFORMANCE RUN | fredrezones55/Qwen3.5-Uncensored-HauhauCS-Aggressive:4bQ4_K_M | PARTIAL | 18.39 tok/s | 5/5 | 1 | verified | INELIGIBLE |
REAL CANARY RUNS
Performance probes only
| Run ID | Model | TTFT | Prompt | Generation | Cold/Warm | Cleanup |
|---|---|---|---|---|---|---|
HB-20260907-C743D6 REAL CANARY RUN | moondream:latestQ4_0 | NOT COLLECTED | 133.06 tok/s | 1000000.00 tok/s · 1-token sample | cold | verified |
FICTIONAL FIXTURES
Mock pipeline records
| Run ID | Model | Suite | Backend | Created | Reproducibility |
|---|---|---|---|---|---|
HB-20260907-DEMO01 MOCK | Model A Q6Q6 (simulated) | HeavryBench Demo v0 | mock 0.1.0 | 2026-09-07 | unverified |
HB-20260907-DEMO02 MOCK | Model A Q4Q4 (simulated) | HeavryBench Demo v0 | mock 0.1.0 | 2026-09-07 | unverified |
HB-20260907-DEMO03 MOCK | Model B Q8Q8 (simulated) | HeavryBench Demo v0 | mock 0.1.0 | 2026-09-07 | unverified |