Loading activity…
@arena
Where AI meets the real world. We measure and advance the frontier of AI through community-driven evaluation. We’re hiring → http://arena.ai/jobs
AI
His early-access testing favored one agent for coding projects, but clear goals and human oversight still mattered.
AI
With both models at max effort, Ember matched Claude Haiku 5.5’s Intelligence Index score at roughly one-third the task cost.
AI
The model supports image inputs, with fine-tuning and inference available at launch and downloadable weights planned for later.
AI
Omni Flash 1.2 was described as out on November 5; a separate post says it offers longer, cheaper videos.
AI
Kimi K3.1-Pro's open-weights announcement links to a Hugging Face URL under moonshotai.
AI
Agent Arena reports that 86% to 96% of agents' tokens are context they have already seen.
AI
Arena says Amazon scientist Ido Rosen discusses evaluation pitfalls and how open-weight models could affect frontier-lab pricing.
YouTubeAI
Agent Arena says the OpenAI model recorded 10.6% net improvement and a $0.62 median cost per task.
AI
Arena says Exo separates an agent’s state, history, secrets and “guardian” from a policy layer that can evolve.
Technology
Google plans to start with a subset of cyber defenders, then expand access through its Fairwind Program.
AI
A post says Carbon scored 88.2% on SEC-bench Pro and is rolling out first to cybersecurity specialists.
AI
Arena says users can now connect GitHub, Gmail and Google Calendar within a single Agent Mode task.
XAI
Artificial Analysis says the model leads four of 10 editing use cases, including live-action film and UI/UX design.
AI
OpenAI says Terra can interpret drawings, act on application screens and draft text.
YouTubeTechnology
Moonshot says both versions are live on its services and API, with open weights planned by Nov. 4.
AI
Kimi claims a 1-million-token context window for Pro and two to three times Pro's speed for Highspeed.
XAI
Arena's October 23 tease quoted a July result for Kimi-K3, not a ranking for the coming model.
AI
Agent Arena says GPT-6 Luna takes three of four optimal spots below $0.10 per task at three reasoning levels.
AI
Artificial Analysis put Ultra fifth on AA-Image-Editing v2.1; Arena ranked it fourth in Multi-Image Edit on October 20.
XAI
A post says one commit references a K3.1 deployment already running on Deepinfra.
