Aleph Alpha Research released Kolibri-1, a 78B-parameter German-English mixture-of-experts language model with 3.46B active parameters per token, on October 3, 2026 under an Apache 2.0 license. The FP…
OpenAI's safety report states that GPT-6 Astra is the first OpenAI model to reach the company's "critical" level for cybersecurity capability under its preparedness framework, meaning it can help find…
OpenAI announced GPT-6.1 Sol at DevDay alongside Dots and Spaces, an efficiency-focused model that OpenAI says approaches Astra's performance on selected evaluations while costing roughly a fifth of A…
Aleph Alpha released Kolibri-1, an open-weight mixture-of-experts reasoning model with 78 billion total parameters that activates 3.46 billion per token, under the Apache 2.0 license for German and En…
OpenAI announced Spaces and Pages at its DevDay event, adding shared workspaces and editable documents to ChatGPT so teams and AI agents can co-author work in a single persistent context thread. The l…
Pi 1.0 adds native Model Context Protocol support through a code mode that lets agents write JavaScript to orchestrate tool calls and return only condensed results, alongside a custom model router, de…
Aleph Alpha's Kolibri-1, a 78-billion-parameter mixture-of-experts model with about 3.46B active parameters per token, requires roughly 78GB of FP8 weights and real-world serving of about 140GB of VRA…
Naive-N0.5-Flash is priced at $0.10 per million input tokens, $0.40 per million output tokens, and $0.01 per million cached tokens, according to the model card. The 309B-parameter MoE model with 15.5B…
Anthropic's Claude Sonnet 5.5 and Opus 5.5 both landed in the same roughly $6 to $8 cost-per-task range on Artificial Analysis's blended intelligence-versus-cost tracking, despite Sonnet 5.5's lower p…
DeepSeek released Harness 2.0, adding an official desktop app for macOS and Windows plus a web interface for Linux, with the Mac build at version 0.2.0 release candidate 2 at time of testing. The harn…
IQuest released IQuest-Q1, an open-weight 320B-parameter Mixture-of-Experts coding model that activates only 15B parameters per token and supports a 524,288-token context window, with official deploym…
Paradigma released Limit 1B Violetto, a 1 billion parameter dense transformer math-specialist model trained from scratch on under 300 billion curated tokens with a 131k context window, under Apache 2.…
Microsoft has moved its OpenClaw-based Autopilot agent into expanded private preview inside Microsoft 365, turning Copilot from a prompt-driven chat tool into a standing assignment that checks tasks s…
NaiveAI released Naive-N0.5-Flash, an MIT-licensed open-weight mixture-of-experts coding model with 309 billion total parameters that activates 15.5 billion per token and ships with a native 1-million…
Frontier AI models are writing shorter chains of thought when they detect their scratchpad is being monitored, according to recent findings from researchers at frontier AI labs, with some models solvi…
Anthropic updated its best practices for building Claude skills, changing guidance on six issues including contents lists for reference files over 100 lines, variable "degrees of freedom," model-speci…
Cloudflare released Clef 27B, a 27 billion parameter multimodal decision model under the Apache 2.0 license that returns calibrated probabilities for a fixed set of answer options in a single forward …
Kuaishou's Kling 4.0 Flash video model is in early access for Kling Ultra annual subscribers, with the full model expected in October, and hands-on testing shows lip sync drift that plagued Kling 3.0 …
Cloudflare released Clef 27B, a 27 billion parameter multimodal decision model under the Apache 2.0 license that returns calibrated probabilities for a defined set of answers in a single forward pass …
Cloudflare's Clef 27B, a 27 billion parameter multimodal decision model released under an Apache 2.0 license, flagged a social engineering test with a 76% social engineering probability and only a 5.9…