A study covered by The Economist found that students using AI for homework raised their scores 18% and cut completion time from 64 to 45 minutes, but scored 20% lower than peers on exams. An ICML 2026…
OpenAI cut API prices for its flagship GPT-5.6 Sol model by more than 20% for three months starting August 21, reducing input tokens from $5 to $4 per million and output tokens from $30 to $20 per mil…
Nvidia reported a 100.00 RHAE score on the ARC-AGI-3 public set using its Agentic Variation Operators (AVO) system, up from Claude Opus 5's 30.16% published score, by wrapping the same model family in…
A new benchmark testing eight headless coding-agent CLIs on a single Python task found that seven of eight agents passed on both runs, but list prices per run varied 18-fold, from $0.0165 (DeepSeek V4…
Nvidia published research on Friday showing that a custom harness with a supervisor component enabled Claude Opus 5 to achieve a 100% score on the ARC-AGI-3 benchmark, compared to 30% without the harn…
NVIDIA's AVO coding-agent system scored 100% on the public ARC-AGI-3 benchmark, completing all 183 levels in 6,624 environment actions, according to NVIDIA AI. The system, powered by Claude Opus 5, us…
An engineer discovered that Claude prompt caching misses its 20-block lookback window in agent loops, causing silent cache misses and 12x cost swings. They detail how parallel tool calls append 18 blo…
Nvidia's AVO agent, powered by Anthropic's Claude Opus 5, achieved a 100% success rate on the ARC-AGI-3 benchmark, completing all 183 levels across 25 environments on August 21 while using 12% fewer a…
DeepSeek, a Chinese AI startup, released DeepSeek-V4-Flash-Vision-Exp, a multimodal AI model that reportedly approaches the performance of Anthropic's Claude Opus 5, intensifying competition in the AI…
Veracode's 2026 GenAI Code Security Report, released July 28, found that AI-generated code passes security checks only 56% of the time across more than 100 models tested over two years, up just one pe…
OpenAI has introduced promotional pricing for its GPT-5.6 series on OpenRouter, offering discounts of up to 50% with prices starting at $0.10 to $2.50 per million input tokens, undercutting Anthropic'…
In a blind comparison, three AI models preferred machine-generated literary passages over works by famous authors more than 90% of the time, with DeepSeek V4 Flash choosing machine text in 94% of deci…
Google's Gemini 3.7 Flash scored 60% on Artificial Analysis' AA-AnalystAgent benchmark, outperforming all other models tested, including Anthropic's Claude Opus 5 at 53.8% and OpenAI's GPT-5.6 Sol at …
JetBrains released Rider 2026.2.1 and ReSharper 2026.2.1, the first minor updates for the 2026.2 release cycle. Rider 2026.2.1 bundles a refactoring-code skill that makes AI agents 83% faster, 64% che…
The developers behind the open-source OmnisBench project have released a benchmark designed to measure the true cost savings of LLM routing. Running 364 coding and math tasks across four models, they …
Anthropic reported degraded performance for Claude Opus 5 and Claude Haiku 4.5, with the incident page offering email and SMS updates for affected users. The status page did not specify the nature or …
Anthropic's beta for mid-conversation tool changes, shipped with Claude Opus 5, fixes a prompt-caching bug that caused cache misses when the `tools` array was modified mid-session, costing up to 87% s…
Z.AI's GLM-5.3 scored 60 on the Artificial Analysis Intelligence Index, tying Kimi K3 for the top spot among open-weights models and sitting three points behind Claude Opus 5's 63, with weights expect…
Anthropic reported that its Claude AI models, including the Mythos Preview and Opus 4.8, successfully designed protein binders against 14 of 15 targets, with 22% to 35% of individual designs binding s…
Anthropic is introducing a staged access program for life-science researchers to use its frontier AI systems, beginning with a restricted cohort of biology researchers for the Mythos 5 model. The comp…