Expired50% confidenceBenchmarkExact time
In tests covering context lengths of 512–4096, increasing the draft length from 4 tokens to 16 tokens resulted in only a 0.2%–1.3% increase in overall inference latency.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/17/2026
First Seen
This claim has expired. It is kept for the record and no longer describes the current state.
Valid until: 10/15/2026(expired)
Sources
Related Claims
Unverified千问3.8 27B的代码方向进步明显大于文本创作，剧本拆解能力相比3.6有提升但幅度有限69% similarUnverifiedAt a draft length of 15 tokens, DSpark's efficiency gains in math, code, and dialogue scenarios reached 30%, 26%, and 22% respectively.64% similarUnverifiedIn a Chinese character count control test requiring exactly 40 characters, Claude Sonnet 4.5 completed exactly 40 characters, while GPT 5.1 wrote 37 characters.61% similarUnverifiedEarly testing on the Opus model showed an exact string recognition rate of 0 out of 15, whereas after the release of Fable 5 the same test yielded 13 out of 15.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/541581API
curl https://kongchang.com/api/v1/knowledge/claims/541581MCP
get_claim(id=541581)