DeepSeek V4 Pro's scores jumped without changing the model at all.
Someone changed how it runs. The claim: V4 was already capable. The gap was never the model. It was the runtime. A developer built JSpace, a wrapper that sits around the model while it works. Same weights.
IPad connected over WiFi AP to the Spark running DeepSeek V4 Flash.
Imagine rolling up to the coffee shop with this naughty little setup. Harness on the iPad time 👀😅 NVIDIAAI vllm_project OpenWebUI Apple
The #1 model on our gateway changed 11 times between Jan 1 and Aug 14.
8 different models held the top spot. Average reign: about three weeks. Feb 13: Opus 4.6 was 78% of all tokens routed. Aug 14: DeepSeek V4 Flash, 32.8%.
First daily drop: DeepSeek V4 Flash leads at $0.03, while GPT-5.6 Sol tops the AAII score at 61.
You have reached the end of the archive
All of deepseek