Unverified50% confidenceBenchmarkExact time
When running Qwen3 27B (MXFP4) at 16 concurrent sessions, aggregate prefill reaches 6,300–6,800 tok/s, and aggregate decode throughput peaks at approximately 900 tok/s.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/11/2026
First Seen
Valid until: 1/9/2027
Sources
Related Entities
Related Claims
UnverifiedOn a single RTX PRO 6000, peak VRAM for Qwen3.8-27B-NVFP4 dropped from 72.9 GB to 40.2 GB, and per-step time was reduced by approximately 11%69% similarUnverifiedUnoptimized baseline token decoding speeds for running Qwen 3 (27B): RTX 5090 ~75, RTX 4090 ~46, RTX 3090 ~40, M5 Max 31.4, DGX Spark 12–17, Ryzen AI Max 11.5 tokens/s.69% 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.68% similarVerifiedQwen3 27B scored 90.3 on LiveCodeBench V6, slightly surpassing Opus 4.6 Max's score of 88.8.68% similarUnverifiedA ~25 GB Q3 model running entirely within the RTX 4090's VRAM (bandwidth ~1 TB/s) theoretically achieves ~40 tokens per second, whereas running it in DDR5 memory (~80 GB/s) drops throughput to 1–3 tokens per second.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/995431API
curl https://kongchang.com/api/v1/knowledge/claims/995431MCP
get_claim(id=995431)