AGI introduces AGI-RUN, an on-device inference engine up to 9.9× faster than SOTA
{"**Who**":"AGI Inc.","**What**":"Introduced AGI-RUN, an on-device inference engine","**How**":"Benchmarked against six models' native engines, on-device SDK startups, and SOTA frameworks on open source datasets","**Why**":"Achieved significant speed improvements, with implications for on-device AI latency and model throughput"}
Published:
Official Source (Official Blog)