Australia open-sources Matilda Jev, a 56.8ms decision model that skips text generation
Med1_Ai · x · 2026-10-06
Australian company Maincode open-sourced Matilda Jev, billed as Australia's first "decision model": no text generation, just calibrated probabilities per option in a single forward pass to cut hallucination and latency. It posted a 56.8ms median latency on a single AMD MI355X across 150,317 evaluation requests, was post-trained end-to-end on local infrastructure as a sovereign AI play, and is available on Hugging Face.
Related event: Australia's Maincode Open-Sources Matilda Jev, Its First Decision Model(2 posts)→
More from Infra
- VC Funding Fell 66% in Last Hike Cycle, Putting ~25% of AI Lab ARR at Risk — menhguin · 2026-10-06
- TensorRT is 206% faster than llama.cpp in this local inference test — kalyan_kpl · 2026-10-06
- Consistent Hashing Explained: Why Modulo Assignment Reassigns Nearly Every Request — _jaydeepkarale · 2026-10-06
- CME launches compute futures as BlackRock's Larry Fink hails 'a new asset class' — Saul_Loveman · 2026-10-06
- Nvidia rethinks AI compute deals, swapping rental guarantees for cloud revenue cuts — rohanpaul_ai · 2026-10-06
- With Nvidia GPU prices so high, Intel cards start looking attractive — lxfater · 2026-10-06