Latest AI Technology

Track the latest breakthroughs and technological advances in AI

69 articles

Anthropic Claude Fable 5.1 Officially Released: 75% Cache Read Cost Reduction, Terminal-Bench-Science Score Doubles, Agentic Workflow Costs Cut by Up to 45%

Anthropic Claude Fable 5.1 released September 1, 2026, with 75% cache read cost reduction saving up to 45% on agentic workflows, Terminal-Bench-Science score jumping from 24.7 to 52.6, and three breaking API changes.

Sep 2, 202611

Anthropic Model Hardware Standard (MHS) Research Preview: AI Agents Directly Control Physical Equipment, QuEra Quantum Laser Success Rate Jumps from 58% to 99.3%

Anthropic launched the Model Hardware Standard (MHS) research preview on August 27, 2026, enabling AI agents to directly control physical equipment like robotic arms and quantum lasers. QuEra's laser success rate jumped from 58% to 99.3%, CMU experiments ran three times faster, marking AI's entry into the physical world.

Sep 1, 202612

Anthropic Claude Opus 5 Officially Released: Matches Fable 5 Performance at Same Price, ARC-AGI-3 Score Triples, 85% Fewer Refusals

Anthropic released Claude Opus 5 on July 24, 2026, positioned as a 'near-frontier' flagship model delivering Fable 5-competitive performance at the same pricing as Opus 4.8 ($5/$25 per million tokens). Opus 5 triples the next-best model's ARC-AGI-3 score, reduces refusals by 85%, and supports a 1-million-token context window, becoming the default model for Claude Max and Claude Pro users.

Aug 31, 202612

Zhipu AI GLM-5.3-Flash Revealed: Mystery 'Ox Alpha' Model Unmasked, 320B MoE Architecture Sets Record 70 Million Daily Requests

On August 26, 2026, Zhipu AI officially confirmed the mystery model 'Ox Alpha' as GLM-5.3-Flash — a 320B parameter Mixture-of-Experts model with a 1M token context window and native multimodal capabilities, released under MIT license, achieving 70 million daily requests during anonymous testing while served entirely on domestic Chinese AI chips.

Aug 30, 202611

Google Gemini 3.7 Flash Officially Released: DeepSWE Coding Score Jumps to 65.3%, Just $0.75 per Million Tokens

Google released Gemini 3.7 Flash on August 13, 2026, with DeepSWE v1.1 coding benchmark scores jumping from 49.0% to 65.3%, supporting 1M token context at just $0.75 per million input tokens, specifically optimized for agentic workflows.

Aug 28, 202611

OpenAI Astra Mathematics Breakthrough: Deep Implications for AI Research and Business

Astra matters not because AI got better at math, but because reasoning may finally become reliable enough to redesign real business workflows.

Aug 22, 202620

Alibaba Qwen3.8-Max: The 2.4-Trillion-Parameter Breakthrough, Pricing, and Enterprise Prospects

Qwen3.8-Max matters less for its parameter count than for how it could reset enterprise AI pricing, deployment, and governance in Asia.

Aug 19, 202616

The 2026 Open-Source AI Model Landscape: Llama, Qwen, DeepSeek and Self-Hosting for Business

In 2026, the open-source AI question is not who has the best model—it’s who understands the cost, compliance, and control boundary best.

Jul 26, 202618

GPT-5: Everything You Need to Know About Capabilities, Pricing, and Business Impact

GPT-5’s real value is not better chat. It is whether it can finish workflows end-to-end with acceptable cost, governance, and ROI.

Jul 1, 202618

PreviousPage 6 / 6