Unverified50% confidenceBenchmarkExact time
在同一张英伟达H100显卡上，DiffusionGemma每秒可生成约1500个token，而上一代自回归模型仅能生成约300个，快了近5倍
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/27/2026
First Seen
Valid until: 11/25/2026
Sources
Related Entities
Related Claims
UnverifiedAlibaba's new-generation Qianwen model has 2.4 trillion parameters, making it the first multimodal large model in the series to exceed one trillion parameters; its code generation accuracy is reportedly 94.7%, an improvement of approximately 11 percentage points over the previous generation.68% similarUnverifiedEarly LLMs could only process approximately 4,000 tokens, while the latest current models already support context lengths of 100,000 or even 1,000,000 tokens.67% similarUnverified报告测出DiffusionGemma平均每步生成token数达到19.74个，而自回归方法每步只能生成1个66% similarUnverifiedThe GPU resources required for a single inference of a hundred-billion-parameter flagship model may be 10 to 20 times higher than those of a ten-billion-parameter model.61% similarUnverified自回归生成机制下，生成长度越长累积计算量越大，生成1000个Token的耗时大约是生成100个Token的10倍61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/807335API
curl https://kongchang.com/api/v1/knowledge/claims/807335MCP
get_claim(id=807335)