REAL LOCAL SUITES · CODING V2 + COMPREHENSIVE · SUITES NOT MIXED
AVAILABLE/PARTIAL PERFORMANCE
OLLAMA LOCAL REGISTRY
moondream:latest
gguf · Q4_0 · 1B
Capability scoreNOT TESTEDReal performance benchmark available
NORMALIZED IDENTITY
Digest-bound metadata
- Registry name
- moondream:latest
- Namespace
- none
- Base name
- moondream
- Tag
- latest
- Family
- phi2
- Parameters
- 1B
- Quantization
- Q4_0
- Disk size
- 1.62 GB
- Architecture
- phi2
- Confidence
- digest bound
- Digest
- 55fc3abd386771e5b5d1bbcc732f3c3f4df6e9f9f08f1131f9cc27ba2d1eec5b
QUEUE POSITION
Not queued
- Queue mode
- PLANNING ONLY
- Item status
- not queued
- Execution enabled
- FALSE
- Attempts
- 0
PLANNED BENCHMARK SUITES
Recorded for a future authorized run
PerformanceChineseEnglishReasoningMathCodingInstruction FollowingStructured OutputAnswer QualityLong Context
IDENTITY RELATIONSHIPS
No relationships detected
No same-digest alias, exact content duplicate, or metadata-matched variant candidate was found.
REAL PERFORMANCE RUN
PARTIAL · HeavryBench Performance v1
- Model
- moondream:latest
- Quantization
- Q4_0
- Hardware
- Apple M4
- Ollama
- 0.30.9
- Prompt version
- Performance Prompt v1
- Methodology
- HeavryBench Performance v1
- Classification
- REAL_PERFORMANCE_RUN
- Leaderboard
- INELIGIBLE
SPEED
Cold and warm measurements
- Cold TTFT
- 2432.82 msrequest → first visible token; may include loading
- Cold load duration
- 2113.50 msseparately reported by Ollama
- Cold generation
- 114.70 tok/s · INSUFFICIENT_SAMPLE
- Cold E2E
- 2527.18 ms
- Warm median TTFT
- 43.52 msresident model: request → first visible token
- Warm median generation
- not collected
- Small prompt median
- not collected
- Medium prompt median
- not collected
STATISTICAL STABILITY
Median is the primary display value
| Metric | Valid/total | Median | Mean | P25 | P75 | Min | Max | Std dev | Outliers |
|---|---|---|---|---|---|---|---|---|---|
| Warm medium generationINSUFFICIENT_SAMPLE | 0/5 | not collected | not collected | not collected | not collected | not collected | not collected | not collected | 0 |
| Warm medium TTFTVALID | 5/5 | 43.52 ms | 48.28 ms | 43.40 ms | 44.23 ms | 42.52 ms | 67.75 ms | 9.75 ms | 1 |
| Small prompt processingNOT_COLLECTED | 0/3 | not collected | not collected | not collected | not collected | not collected | not collected | not collected | 0 |
| Medium prompt processingNOT_COLLECTED | 0/3 | not collected | not collected | not collected | not collected | not collected | not collected | not collected | 0 |
MEMORY
Passive observations with explicit sources
| Point | System free | System available | System used | Process RAM | Ollama residency VRAM |
|---|---|---|---|---|---|
| Before | 0.07 GiBsystem_api | not collectednot_collected | 15.93 GiBsystem_api | not collectednot_collected | 0.00 GiBollama_api · residency, not peak |
| After load | 0.09 GiBsystem_api | not collectednot_collected | 15.91 GiBsystem_api | not collectednot_collected | 1.19 GiBollama_api · residency, not peak |
| After unload | 0.23 GiBsystem_api | not collectednot_collected | 15.77 GiBsystem_api | not collectednot_collected | 0.00 GiBollama_api · residency, not peak |
RAW TRIALS
14 retained trials
Expand every cold, warm, generation, and prompt-processing trial
| Trial | Kind | Profile | State | Tokens | TTFT | Prompt tok/s | Generation tok/s | Validity | Outlier |
|---|---|---|---|---|---|---|---|---|---|
| trial-01 | generation | MEDIUM | cold | 11 / 128 | 2432.82 ms | not collected | 114.70 tok/s | INSUFFICIENT_SAMPLE | no |
| trial-02 | generation | MEDIUM | warm | 11 / 128 | 67.75 ms | not collected | 114.11 tok/s | INSUFFICIENT_SAMPLE | no |
| trial-03 | generation | MEDIUM | warm | 11 / 128 | 44.23 ms | not collected | 113.61 tok/s | INSUFFICIENT_SAMPLE | no |
| trial-04 | generation | MEDIUM | warm | 11 / 128 | 43.52 ms | not collected | 108.52 tok/s | INSUFFICIENT_SAMPLE | no |
| trial-05 | generation | MEDIUM | warm | 11 / 128 | 43.40 ms | not collected | 114.86 tok/s | INSUFFICIENT_SAMPLE | no |
| trial-06 | generation | MEDIUM | warm | 11 / 128 | 42.52 ms | not collected | 105.20 tok/s | INSUFFICIENT_SAMPLE | no |
| trial-07 | generation | SHORT | warm | 11 / 32 | 44.30 ms | not collected | 111.42 tok/s | INSUFFICIENT_SAMPLE | no |
| trial-08 | generation | LONG | warm | 11 / 256 | 43.13 ms | not collected | 114.02 tok/s | INSUFFICIENT_SAMPLE | no |
| trial-09 | prompt_processing | PROMPT_SMALL | warm | 4 / 4 | 139.64 ms | not collected | 126.96 tok/s | VALID | no |
| trial-10 | prompt_processing | PROMPT_SMALL | warm | 4 / 4 | 45.25 ms | not collected | 121.32 tok/s | VALID | no |
| trial-11 | prompt_processing | PROMPT_SMALL | warm | 4 / 4 | 45.44 ms | not collected | 135.30 tok/s | VALID | no |
| trial-12 | prompt_processing | PROMPT_MEDIUM | warm | 4 / 4 | 762.00 ms | not collected | 123.90 tok/s | VALID | no |
| trial-13 | prompt_processing | PROMPT_MEDIUM | warm | 4 / 4 | 47.96 ms | not collected | 125.16 tok/s | VALID | no |
| trial-14 | prompt_processing | PROMPT_MEDIUM | warm | 4 / 4 | 48.23 ms | not collected | 121.71 tok/s | VALID | no |
SAFETY & REPRODUCIBILITY
Cleanup verified
- Before / during / after
- 0 / 1 / 0
- Target absent after
- true
- Digest matched
- true
- Reproducibility
- partial
CAPABILITY: NOT TESTED
HeavryBench Score and all Chinese, Reasoning, Math, Coding, Knowledge, Instruction Following, and Long Context fields remain null.