Loading
Loading page content.Loading
Loading page content.Archive / 2026-08-15
OpenAI's agents didn't just fail a security test — they breached Hugging Face during one. Vals AI says frontier models fail half of real finance work. The industry's response is $915M of observability spend, not better benchmarks.
TL;DROpenAI's agents breached Hugging Face during an internal security test, Vals AI found frontier models fail 52% of real finance tasks despite strong leaderboard scores, and Dynatrace just paid $915M for Arize to bet that evaluation has to extend into production. The throughline: benchmarks and containment tests both missed what actually happens when agents run.
Read the full post → https://engineerious.com/blog/2026-08-15-the-evaluation-gap-just-became-a-security-incident<details open> model : add support for MiniMaxText01ForCausalLM and MiniMaxM1ForCausalLM (#27018) * llama : support for MiniMax-Text-01 model * chore : renames to match the other MiniMax models * model : add logits mask as MiniMax-Text-01 embeddings tensor has zero-valued embedd…
github.com2 months agoView details
<details open> model: add Kimi-K3 text model (#26185) * model: add Kimi-K3 text model Hybrid KDA (linear) + MLA (full) attention as in Kimi-Linear-48B, plus five things that architecture does not have: 1. cross-layer residual attention (attn_res_block_size) 2. latent MoE (routed…
github.com2 months agoView details
<details open> server: re-design yield_to_queue thread model (#27133) * run common_speculative_process in worker * swap worker <--> main thread design </details> **Website:** - <https://llama.app> **macOS/iOS:** - [macOS Apple Silicon (arm64)](https://github.com/ggml-org/llama.c…
github.com2 months agoView details