Unverified50% confidenceFactExact time
The newly added ultra-fast mode increases generation speed by up to 8x, capable of outputting approximately 300 tokens per second.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/1/2026
First Seen
Valid until: 12/30/2026
Sources
Related Entities
Related Claims
UnverifiedFast Mode delivers 1.5× speed but consumes 2.5× quota.68% similarUnverifiedAt approximately 52 concurrent connections, the system is essentially 100% saturated, generating a maximum of approximately 704 tokens per second, but with latency spiking to 64 seconds.68% similarUnverifiedIn API scenarios, Ultrafast's token generation speed is more than 7 times that of the Standard Tier.66% similarUnverifiedWhen running IQ3_XXS quantized models with llama.cpp, the output speed is only 15 tok/s with prompt processing at 100–120 tok/s, while a self-developed engine brings the same quantization up to approximately 65 tok/s output and 430 tok/s prompt processing, representing close to a 4x performance improvement.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/966072API
curl https://kongchang.com/api/v1/knowledge/claims/966072MCP
get_claim(id=966072)