Monitoring of the deepseek/deepseek-v4-flash endpoint on open-inference/fp8 was retired with no border inputs found since 2026-09-24, and its logprob tracking stalled after recent queries failed, acco…
A financially motivated threat actor used three open source AI harnesses — Strix for vulnerability search, Cairn for autonomous exploitation, and Hermes for orchestration — to breach hundreds of onlin…
AgentsView launched an installable auto-push service that mirrors each machine's AI coding session archive into PostgreSQL, with a read-only server presenting the merged fleet. The tool imports record…
PixelCrew, an AI design tool from GitLab founder Sid Sijbrandij, launched in free alpha, running a sequential crew of specialized agents that turn a written brief into production-quality design output…
A developer built a pull-request review gate on top of TypeSafe's Jev, a "System One" model released September 15th, 2026 that returns probabilities rather than generated text. The app sends a state o…
OpenRouter launched a Batch API that lets providers complete bundled inference requests anytime within a 24-hour window in exchange for charging roughly 50% or less of normal per-token pricing, availa…
A developer surveyed five MCP gateway solutions — Bifrost, OpenRouter, Cloudflare AI Gateway, Kong AI Gateway, and LiteLLM — for managing Model Context Protocol servers at enterprise scale in 2026. Th…
Xiaomi Corp. released and open-sourced the MiMo-V2.6 series of natively omnimodal generative AI models, comprising the flagship MiMo-V2.6-Pro, a smaller Flash variant, and a Pro-UltraSpeed variant tha…
A developer published a beginner tutorial for building an end-to-end multi-agent RAG legal assistant tailored to UAE Federal Law, using LangGraph, FastAPI, Streamlit, Pinecone, and OpenRouter. The arc…
A test of a small prompt edit on 100 single-call tool-calling cases from UC Berkeley's BFCL V4 benchmark found the edit reduced invented optional arguments from 14 to 7 across three runs but produced …
TypeSafe removed the waitlist for its Jev model on Sep 20, opening registration with $5 in free credit (roughly 120 million tokens). Jev returns typed decisions — Choice, Score, and Noul — with calibr…
AgentSky published a hands-on comparison of Jev Ultrafast (running Jev 1.13) against Codex (running GPT-5.6 Sol at medium thinking level) on a Google Flights search task, with both agents starting fro…
A developer built a working implementation of the Assess-Decide-Do framework using JEV, a new System One AI model that outputs no text and instead answers only through three primitives: noul (yes/no),…
Inception Labs' Mercury 2.5 is the fastest LLM available via API as of September 2026, posting 1,107 tokens per second on vendor benchmarks and 440 tok/s P50 at 1.17s latency in OpenRouter telemetry, …
Xiaomi disclosed that the final reinforcement-learning run for its MiMo-V2.6 models, released September 22, 2026, cost about $850,000 for the Flash model and about $2.62 million for Pro, covering unde…
TypeSafe released jev-graph-search 0.2.2, a CLI that lets an AI agent retrieve and rank notes from local Obsidian vaults and Logseq Markdown graphs, with an optional JSON snapshot. On an 8,851-node ta…
Xiaomi released and open-sourced the MiMo-V2.6 series, two natively omnimodal models for coding, visual tasks and computer use, with MiMo-V2.6-Pro scoring 46.32 on the Artificial Analysis Intelligence…
A developer used GPT 5.6 Luna via Simon Willison's `llm` CLI tool to label git commits as "maintenance" or "new development", reporting that the model matched their own manual labels across the entire…
Treasury Secretary Scott Bessent said the US has proposed a "notification mechanism" to flag AI incidents affecting national security ahead of Chinese President Xi Jinping's scheduled White House visi…
Xiaomi released and open-sourced its MiMo-V2.6 series, publishing Hugging Face repositories for two omnimodal models, MiMo-V2.6-Pro and MiMo-V2.6-Flash, along with a technical report, deployment instr…