Unverified50% confidenceBenchmarkExact time
Moondream 0.5b takes about 7.6 seconds per new frame on a 4-core CPU, faster than Qwen3-VL's 18 seconds and gemma3:4b's 17 seconds.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/19/2026
First Seen
Valid until: 1/17/2027
Sources
Related Entities
Related Claims
UnverifiedThe UP host tested MiniCPM-2B on a computer with 4G VRAM and achieved a generation speed of 57 tokens per second, with a first-token latency of only 0.22 seconds.76% similarUnverifiedOn a 4-core CPU, Qwen3-VL 2B-instruct never wrongly answered 'yes' for things that do not exist in the 128 annotated images, but each new frame takes about 18 seconds to read.72% similarUnverifiedJev's reported server time is approximately 88 milliseconds per classification, while running GLiNER 2.5 small locally on a MacBook M4 Pro GPU is approximately 8 times faster.71% similarUnverifiedWhen generating a 5-second 768P video with Fast Cache enabled, the M3 Ultra took 37 minutes and 14 seconds, while the M5 Ultra took 9 minutes and 24 seconds, nearly 4 times faster.70% similarVerifiedIn benchmark testing, Qwen3 27B achieved a generation speed of approximately 40-something tokens per second on 4 modified 2080 Ti cards without NVLink enabled.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1071201API
curl https://kongchang.com/api/v1/knowledge/claims/1071201MCP
get_claim(id=1071201)