Unverified50% confidenceFactExact time
Inference-enhanced models such as DeepSeek-R1 generate a large number of intermediate reasoning tokens before producing the final answer, and these thinking tokens are counted as output token costs.
1
Sources
50%
Confidence
Long-term
Relevance
8/30/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedDeepSeek-R1 用特殊标签（如 <think>...</think>）在输出中显式区分思考过程与最终答案79% similarUnverifiedThe release of DeepSeek-R1 sparked widespread industry discussion about the inference-time compute paradigm.75% similarUnverifiedDeepSeek V3 and R1, leveraging the MoE architecture, compress inference costs to a fraction of those of traditional dense models while maintaining top-tier performance.74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/825677API
curl https://kongchang.com/api/v1/knowledge/claims/825677MCP
get_claim(id=825677)