Unverified50% confidenceBenchmarkExact time
Under the 1024-token limit, 49 of Haiku 5.5's responses (2.2%) were truncated, which never happened with Haiku 4.5; 6 of these came from the prompt "Explain the word 'time' in exactly 50 words".
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/18/2026
First Seen
Valid until: 1/16/2027
Sources
Related Entities
Related Claims
UnverifiedThe cache hit rate for exact SHA256 hash caching of prompts is often less than 4%, because an extra space or punctuation mark will invalidate the cache.70% similarUnverifiedAfter enabling reasoning mode, Qwen3.8-27B's bug detection rate dropped to 6/20, and 12 of 40 answers could not be parsed as JSON because they exhausted the 12,000-token output limit; Ornith 1.5's detection rate remained 8/20, with 13/40 answers unparseable, but it took about half the time of Qwen.59% similarUnverifiedIn a binary classification task, if the positive-to-negative sample ratio is 6:4, then a baseline that always outputs "positive" scores 60 points, and an agent scoring 65 has only 5 points of real signal.58% similarUnverifiedFor the same failing test fix, Haiku costs only about one-sixth of a cent—17 times cheaper than Sonnet—and completes in 6 seconds (compared to 10 seconds for Sonnet).57% similarUnverifiedSmolLM2 360M fine-tuned with LoRA achieves an F0.5 of 0.663 on the grammar correction task, with a 20.1% false-edit rate.57% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1061250API
curl https://kongchang.com/api/v1/knowledge/claims/1061250MCP
get_claim(id=1061250)