Unverified50% confidenceEventExact time
A developer ran the 2.8 trillion-parameter Kimi K3 model on a single MacBook Pro at approximately 1 token/second.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/11/2026
First Seen
Valid until: 12/10/2026
Sources
Related Entities
Related Claims
UnverifiedKimi K3 requires approximately 1.5 TB of memory under Q4 quantization, potentially meaning four Macs would need to be linked together.68% similarUnverifiedIn a controlled test, the 35B model ran at approximately 100 tokens/second (Mac Studio), roughly five times the 9B model's 16 tokens/second67% similarUnverifiedRed Note AI Lab launched a preview version of DOS 3 Note, a MoE model with 28 billion total parameters, only 1.6 billion activated parameters, supporting 512K ultra-long context and multimodality.66% similarVerifiedKimi K3 supports an ultra-long context window of 1 million tokens.66% similarUnverifiedKimi announced the full public release of K2.7 Code High Speed, with an output speed of approximately 180 tokens per second in typical programming scenarios—about 5 to 6 times that of the standard version—at 3 times the credit consumption of the standard version.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/904679API
curl https://kongchang.com/api/v1/knowledge/claims/904679MCP
get_claim(id=904679)