Unverified50% confidenceBenchmarkExact time
The 11 most well-known writing tics lost 53% of their strength from Opus 4 to Opus 5.5, while over the same period the total number of tics dropped from 3,746 to 2,548, a decrease of only 32%.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/16/2026
First Seen
Valid until: 1/14/2027
Sources
Related Entities
Related Claims
UnverifiedGraphite's research found that there are 2,548 writing tics in Claude Opus 5.5, only 4% fewer than the 2,666 in Opus 5.67% similarUnverifiedExperience internalization fine-tuning improved Qwen3-8B-27B's one-shot score from 37.07 to 48.22, while reducing generated tokens from 6.41 million to 570,000 (an 11-fold reduction).63% similarUnverifiedOn models of 4B and below, the normal-style prompt scored highest, with the plain and florid styles both dropping about 8 points, but all confidence intervals crossed zero, so the differences cannot be strictly proven significant.61% similarUnverifiedTEMPEST outperforms the best classical model by 17.9 percentage points and the strongest triplet-loss baseline by 58.4 percentage points.60% similarUnverifiedWith the addition of an on-chip language model with 1.3 million parameters, Oído's WER drops to 3.33% / 7.15%.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1041464API
curl https://kongchang.com/api/v1/knowledge/claims/1041464MCP
get_claim(id=1041464)