TECHNOLOGYJun 8 · 15:27 UTCHACKER NEWSgainsurier

MiMo-v2.5-Pro-UltraSpeed: 1T model with 1000 tokens per second

MiMo-v2.5-Pro-UltraSpeed is a 1T model capable of processing 1000 tokens per second. The article, hosted on Xiaomi's blog, has received 66 points and 14 comments on Hacker News.

Nexus surfaces and summarizes. The full story lives at the source.

Mentioned

XiaomiorganizationMiMo-v2.5-Pro-UltraSpeedtopicHacker Newsorganization

Spot something wrong with this article?Report a problem →

Forward this

Related Signal

Adjacent reporting

TECHNOLOGYApr 20 · 18:46 UTC
We got 207 tok/s with Qwen3.5-27B on an RTX 3090
TECHNOLOGYApr 24 · 03:07 UTC
DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence
TECHNOLOGYMay 29 · 09:47 UTC
Real-time LLM Inference on Standard GPUs: 3k tokens/s per request
TECHNOLOGYMay 6 · 15:44 UTC
Google's Gemma 4 AI models get 3x speed boost by predicting future tokens
TECHNOLOGYApr 23 · 19:25 UTC
OpenAI Releases GPT-5.5: Faster, Smarter—And Pricier
TECHNOLOGYMay 15 · 22:38 UTC
Orthrus-Qwen3: up to 7.8×tokens/forward on Qwen3, identical output distribution