TOOLS23 SEPT 2026 · 09:34 UTC● LIVE COVERAGE

UK AISI and EvalEval Publish Reproducible Benchmark Results

UK AI Security Institute and the EvalEval Coalition have published reproducible benchmark results, releasing Evaluation Cards for five core benchmarks-HealthBench, FrontierMath, Humanity's Last Exam, SWE-Bench Pro and Terminal-Bench 2.0-tested on six frontier LLMs and two cyber-focused evaluations, using the Every Eval Ever schema to ensure transparency.

EDITORIAL DESK

UK AISI and EvalEval Publish Reproducible Benchmark Results — UK AI Security Institute and the EvalEval Coalition have published reproducible benchmark results, releasing Evaluation …

Latest dispatches

UPDATED 09:34 UTCALL COVERAGE →

MODELS09:29 UTC

New AI Models Promise Lower Costs for Developers

Anthropic’s Opus 5.5 and OpenAI’s GPT-6 Sol and Luna models lower token prices-Opus 5.5 at $4 input and $20 ou…

EDITORIAL DESK

INDUSTRY09:23 UTC

Microsoft takedown of AI-driven scam service that hit 12,000 accounts

Microsoft announced on September 22, 2026 that it helped dismantle the subscription-based AI-driven fraud plat…

EDITORIAL DESK

TOOLS09:21 UTC

Rabbit releases cloud-based AI agent that works on any PC

Rabbit has launched a cloud-based AI agent called OS3 that runs on any Windows, macOS, or Linux PC without ded…

EDITORIAL DESK

TOOLS09:20 UTC

UK AISI and EvalEval Share Reproducible AI Benchmark Results

UK Artificial Intelligence Safety Institute, working with the EvalEval platform, has published verified result…

EDITORIAL DESK

The briefing wire

REAL-TIME DESK · ALL TIMES UTCOPEN FULL WIRE →

09:29MODELSNew AI Models Promise Lower Costs for DevelopersEDITORIAL DESK 09:23INDUSTRYMicrosoft takedown of AI-driven scam service that hit 12,000 accountsEDITORIAL DESK 09:21TOOLSRabbit releases cloud-based AI agent that works on any PCEDITORIAL DESK 09:20TOOLSUK AISI and EvalEval Share Reproducible AI Benchmark ResultsEDITORIAL DESK 09:18MODELSParallel halves research time and cost with GPT-6 AstraEDITORIAL DESK