Unverified50% confidenceBenchmarkExact time
In Ziskind's hands-on tests, the M5 Ultra achieved three- to four-fold prefill gains on long prompts, with only 1.6x on short prompts; MacStories measured about 2.5x on a 16,000-token prompt, with time to first token dropping from 13.9 seconds to 5.6 seconds.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/14/2026
First Seen
Valid until: 2/12/2027
Sources
Related Entities
Related Claims
UnverifiedSpeculative decoding can make a certain number-intensive test about 5 times faster on an M4 Max.68% similarUnverifiedQwen 2.5 0.5B在WebGPU下达到37-40 tokens/s,中位回复耗时约1.8秒,是三个测试模型中最慢的。68% similarUnverifiedQwen 3.8 has a base speed of 22 tokens/s on the M5 Max, which can be boosted to 34 or even 56 tokens/s with MTP enabled.66% similarUnverifiedIn a model comparison test, qwen2.5:3b correctly understood pace, produced 0 hallucinated place names, had a length error of 0.3%, peak memory of 259 MB, and took 106 seconds.66% similarUnverifiedFor Gemini 4 Argon, the maximum output for synchronous requests is 65,536 tokens, and the maximum output for long decode is 1,048,576 minus the input tokens.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1169140API
curl https://kongchang.com/api/v1/knowledge/claims/1169140MCP
get_claim(id=1169140)