Unverified50% confidenceBenchmarkExact time
研究数据显示，在复杂代码生成任务中，将thinking budget从1024 tokens提升到8192 tokens可以带来15-30%的准确率提升，但推理成本增加4-8倍
1
Sources
50%
Confidence
Long-term
Relevance
8/25/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn Max Reasoning mode, Soul consumes more tokens in exchange for better output, with the reasoning budget increasing from approximately 768 to nearly 1000.72% similarUnverifiedA 54% improvement in token efficiency means that the number of tokens consumed to complete the same programming task is reduced by nearly half.66% similarUnverifiedThe price of output tokens is typically 4–6 times that of input tokens, stemming from the computational differences between the prefill and decode phases, with GPU utilization during the decode phase potentially being only 10–20% of that during prefill65% similarUnverifiedXHigh 档位推理 Token 消耗可能达到 High 档位的 3-5 倍，但准确率边际提升通常仅有 5-15 个百分点65% similarVerifiedWhen the draft model prediction accuracy of Speculative Decoding reaches 60–80%, the overall inference speed can be improved by 2–3 times.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/798417API
curl https://kongchang.com/api/v1/knowledge/claims/798417MCP
get_claim(id=798417)