Every local and reasonable model is bad at that. It’s OpenAI’s own benchmark, so take it with a grain of salt, but “GPT-6 Astra achieved 100.0% accuracy on contexts from 256K to 512K tokens and 96.3% accuracy from 512K to 1M tokens on OpenAI's 8-needle long-context benchmark”.
Depends on how unique the needles are. If they're fairly unique then even Qwen 3.5 27B @ Q5 with 128k context and Q4 KV cache has perfect recall of at least up to 5 needles, according to my tests.
But I'm sure if you make them less unique they'll start to struggle.
it is not about simple or complex circuit. non-linear process creates complicate behavior. a nonlinear process can be very simple, and the output can be very complicated.
99% of prototype boards I've done involve a fine pitch BGA for which via-in-pad is fairly required. This is for prototypes of the product hardware, and doesn't include boards for test hardware etc which is more flexible.
Sorry about that, we went viral in various places this week and demand has been through the roof. We are working round the clock to get more capacity online.
this is not right. vllm can easily load safetensors model at >5GB/s if not faster when setup right. did you make the compile cache persist? if you use docker, you should bind mount the kernel compile cache, so they don't need to be recompiled each time vllm restart.
reply