Unverified50% confidenceCautionExact time
Running this engine requires a Blackwell GPU (the kernels use the NVFP4 hardware path), about 70GB of space on fast NVMe, and 32GB of RAM.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/17/2026
First Seen
Valid until: 1/15/2027
Sources
Related Entities
Related Claims
UnverifiedOn the test machine, FastGPT's memory usage at idle is about 2.1 GB, a VPS with 2GB RAM would struggle even to start, and a machine with at least 4GB of memory is recommended.70% similarUnverifiedSGLang using official NVFP4 weights with D-Spark speculative decoding achieved an average of approximately 173 tokens/second on Blackwell GPUs, approaching 200 for short tasks, while loading a 262k context window.69% similarUnverifiedOn an 8 vCPU, 16GB memory (about 4GB free during testing), GPU-less Linux machine, the 1.7-billion-parameter qwen3:1.7b model could not drive the Anatomy Skill to complete a qualified multi-step workflow.69% similarUnverifiedDuring runtime, GPU utilization reaches 100%, RAM usage is approximately 181 GB, and SoC power consumption is approximately 147 watts.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1053047API
curl https://kongchang.com/api/v1/knowledge/claims/1053047MCP
get_claim(id=1053047)