The new 200-task MKA benchmark, built with Artificial Analysis, evaluates agents on visually complex work across 10 verticals and 83 occupations.
The new top-level domains are available now through an Early Access Program with fees that decrease daily until public launch on November 17, 2026.
The model targets sustained agentic workflows and complex multi-step problems in software engineering and scientific research.
Mitra-v2, a second-generation tabular foundation model, is pretrained on tables with 10x more rows and 3x more columns than its predecessor, leading TabArena and ScoringBench.
A steering vector derived from the difference between loose and strict reward checkpoints allows extrapolation to reduce resource-limit failures from 14.5% to 8.0%.
Egzakta releases Kvark Sparkle 1.1, an Apache 2.0 fine-tune of Qwen3.8-27B that correctly declines answers not found in documents 91.5% of the time.
The Apache-2.0 licensed model family offers three builds, with the Flash variant achieving 0.31 s first audio latency and supporting 40 concurrent users per 80 GB GPU.
A new benchmark of 202 real-world tasks across 13 languages finds Opus 5.5 maintains performance on functional languages while open-weight models like DeepSeek, GLM, and Qwen fall behind.
The open-source framework provides a unified catalog and execution pipeline to separate execution readiness from evaluative claims across video, 3D, and 4D systems.
Anthropic launched a read-only Personal Finance beta and introduced invisible text watermarks for specific models, while a METR report suggests AI is accelerating Anthropic's R&D by roughly twice the previous rate.
The partnership brings Adobe's CX Enterprise Coworker agent to Gemini Enterprise and enables near-real-time Photoshop edits via DeepMind collaboration.
Foxclaw is an open-source agent framework that treats memory as a runtime, achieving 1.0 flag_recall in offline drift detection evaluations.
A November 2026 scan of 277 Model Context Protocol servers found that 55 respond without authentication, including 6 critical servers exposing write-capable tools.
Systematic evaluation across 16 trajectories shows WSP enables substantial localized gains but suffers from inconsistent cross-task generalization and validation reliability.
WAV preserves all audio data for editing but is roughly 11 times larger than a 128 kbps MP3, which is better for sharing and streaming.
Scribe v3 Live reduces realtime semantic errors by 43% and adds speaker diarization, while Scribe v3 improves batch accuracy in every tested language.
The free, in-person events, held during National Family Caregivers Month, are designed to help caregivers use ChatGPT for scheduling, resource navigation, and peer support.
The multi-LLM agentic framework generated over six million lines of Lean code to mathematically prove the security of the actual Signal code in four weeks.
The model uses a single network to generate all hierarchical levels (4³, 16³, 64³) from a single image input.
The framework focuses on correct training by limiting options to SFT, DPO, GRPO, OPD, and SDFT, and includes a feature to visualize which tokens are included in the loss.
The experiment cost $840 in API fees, with a rough estimate that human time per bug dropped from two days to a few hours.
Businesses can claim their profile via Business Hub and launch Business Agents for eligible ecommerce and online service firms in the US.
New ThinkYoung research cited in the report shows a significant gap between student AI usage and teacher guidance in EMEA education systems.
The airline cut agent handling time by 10% and tripled Flight Recommender usage by integrating OpenAI models into customer support and internal workflows.
AI Briefing