Foundation Models - News Articles

OpenAI Discounts Spark Massive Usage Surge, With Luna Jumping 13.8x and Nearly a Third of New Users Sticking Around After Deals Expire

Aug 28, 2026

OpenRouter Blog

OpenAI's limited-time discounts on its Terra and Luna models trigger explosive usage growth — Luna surging 13.8x — while stealing market share from rival AI labs, and nearly a third of new users keep paying full price after the deals expire.

Foundation Models Business

Google DeepMind Launches World's First Double-Blind AI Evaluation Using Cryptographic Technology to Ensure Testing Integrity

Aug 28, 2026

Google DeepMind

Google DeepMind launches the world's first double-blind AI evaluation using cryptographic technology, partnering with global safety organizations to test its Gemini Flash Lite model against confidential benchmarks in a secure environment where neither party can access the other's sensitive data, setting a new standard for trustworthy AI oversight.

AI Safety Foundation Models Security

Cohere Launches Parse, a Vision Language Model That Outperforms Rivals at $1.50 Per 1,000 Pages

Aug 28, 2026

Cohere

Cohere launches Parse, a vision language model that converts complex enterprise documents into structured data, outperforming rivals like Mistral OCR 4, AWS Textract, and Google Document AI at just $1.50 per 1,000 pages, now available via the Cohere API, Microsoft Foundry, and AWS SageMaker.

Enterprise Foundation Models Data & Analytics

OpenAI Codex Update Adds Persistent Reasoning Support But Raises Compatibility Concerns for Custom Providers

Aug 28, 2026

GitHub

OpenAI Codex receives a major update adding persistent reasoning support across its protocol and SDK, but a newly flagged compatibility issue warns that custom provider configurations using a 'persistent' effort value may be silently rewritten to 'disabled', potentially breaking non-OpenAI integrations.

Coding & Developer Tools Foundation Models

Small AI Model Beats GPT-5 and Claude on SQL Benchmark at a Fraction of the Cost

Aug 28, 2026

Thinking Machines Lab

A small AI model called ReViSQL-K2.6, trained with reinforcement learning on a cleaned dataset, achieves human-level 92.97% accuracy on a text-to-SQL benchmark, outperforming GPT-5 and Claude at just 12–15% of their cost.

Coding & Developer Tools Foundation Models Research

Google Launches Gemini Omni 1.1 Flash With 4K Upscaling and 40-Second Scene Extension for Developers

Aug 28, 2026

Google

Google launches Gemini Omni 1.1 Flash, a powerful generative video model giving developers 4K upscaling, 40-second scene extension, and 60% faster draft previews at one-third the cost, available now via Google AI Studio with early adopters like Adobe and Figma already on board.

Image & Video Coding & Developer Tools Foundation Models

Salesforce and Anthropic Launch 'Claudeforce' Plugin, Sending Salesforce Shares Surging 12%

Aug 27, 2026

CNBC

Salesforce and Anthropic unveil 'Claudeforce,' a groundbreaking plugin embedding Salesforce data directly into Claude with 37 pre-built sales skills, marking the first time Salesforce has attached its iconic 'force' suffix to another company's product, sending Salesforce shares surging 12% in extended trading.

Enterprise Business Agents Foundation Models

Alibaba Unveils Qwen4 Architecture Preview With Radical Efficiency Gains Amid Export Control Pressures and EU Legal Uncertainty

Aug 27, 2026

TNW | Artificial-intelligence

Alibaba unveils Qwen3.8-Flash-Next, a preview of its next-gen Qwen4 architecture boasting 125B parameters but activating only 6B per token, slashing compute costs by roughly a third while cleverly sidestepping US chip export restrictions — though its commercial licensing terms may put European users at legal risk under the EU AI …

Foundation Models Geopolitics Regulation Open Source

Tencent Releases WeMM-Embedding Models, Claiming Top Scores on Multimodal Benchmarks

Aug 27, 2026

GitHub

Tencent's WeChat Vision Team releases WeMM-Embedding, a family of multimodal AI models in 2B, 4B, and 9B sizes that claim top scores on MMEB-v2 and MMEB-v3 benchmarks, supporting text, images, videos, and documents, now publicly available on Hugging Face under Apache License 2.0.

Foundation Models Open Source Computer Vision

Google Launches Gemini 3.5 Transcribe With 70% Faster Transcription and Support for 85+ Languages

Aug 27, 2026

Google

Google launches Gemini 3.5 Transcribe, its most advanced speech-to-text model yet, delivering a 70% faster transcription speed, a Word Error Rate as low as 2.6%, and support for 85+ languages — now available in public preview for developers and enterprises.

Audio & Voice Foundation Models Coding & Developer Tools