Unverified50% confidenceBenchmarkExact time
On a Korean NPU server (8 cards, 32 chips in total), VKORE raises the single-user response of the Gemma 4 26B model from 14.1 tokens/s to 53.9 tokens/s, about 3.8 times.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/13/2026
First Seen
Valid until: 10/27/2026
Sources
Related Entities
Related Claims
UnverifiedReal-world testing with the 284-billion-parameter DeepSeek V4 Flash shows token generation rising from 37 to 53 tokens/s (approximately 1.5x), and Prompt Processing jumping from 483 to 1,485 tokens/s (approximately three times).73% similarUnverifiedWhen running Qwen3-Next-Flash (Autoround W4A16 quantization), the prefill speed was approximately 1.3k tokens/s, and the generation speed reached 70 tokens/s (code) / 60 tokens/s (prose), with support for 128k+ long context.69% 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% similarUnverifiedLocally running open-source LLM models typically have a context capacity of only 4K to 32K tokens, limited by hardware memory.68% similarUnverifiedDeveloper Primitive reported: 91.2 tokens/s with speculative decoding off, reaching 142.6 tokens/s after enabling 3 speculative tokens, approximately a 56% speedup under the same configuration.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1013617API
curl https://kongchang.com/api/v1/knowledge/claims/1013617MCP
get_claim(id=1013617)