Unverified50% confidenceBenchmarkExact time
In a token growth test of a real 10-turn task, the context grew from 1450 tokens in turn 1 to 28400 tokens in turn 4, and then to 89600 tokens in turn 10.
1
Sources
50%
Confidence
Long-term
Relevance
10/16/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn speculative sampling, when the acceptance rate of candidate tokens is high, the effective number of output tokens per forward pass can reach 2–4, thereby improving the overall generation speed by 1.5x–3x.74% similarUnverifiedAccording to Artificial Analysis testing, Grok 4.6 consumes approximately 38k tokens per task at the MXX High tier, while Grok 4.7 surges to 81k tokens.70% similarUnverifiedUnder conditions of 2944 input tokens and 1024 output tokens, normal mode reaches 29.9 tokens per second, which improves to 48.4 tokens/s after enabling MTP2.68% similarUnverified父子分块策略用小块（128~256 Token）做精准检索，命中后返回其所属大块（512~1024 Token）作为上下文67% similarUnverified根据Artificial Analysis数据，Fable完成单个任务约用33k token，Opus 5约用37k token67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1044023API
curl https://kongchang.com/api/v1/knowledge/claims/1044023MCP
get_claim(id=1044023)