Adding Benchmaxxer Repellant to the Open ASR Leaderboard

17. May 2026
AI Models

This post reports on anti-benchmaxxer measures introduced to the open ASR leaderboard, featuring updates from contributors including Eric Bezzam, Steven Zheng, Eustache Le Bihan, Sergio Bruccoleri, and Jeanine Sinanan-Singh from the Appen AI Research team, linked to the Hugging Face platform.

Share on:

vLLM V0 to V1: Correctness Before Corrections in Reinforcement Learning

17. May 2026
AI Models

ServiceNow AI resolved a critical train-inference mismatch problem during migration from vLLM V0 to V1, where precise log-probability calculations are essential for consistent reinforcement learning training dynamics.

Share on:

EMO: Pretrained Mixture-of-Experts Model for Emergent Modularity

17. May 2026
AI Models

Allen Institute introduces EMO – a Mixture-of-Experts model in which modular structures develop naturally from data without predefined specifications. The model, code, and technical documentation are freely available.

Share on:

Building Blocks for Foundation Model Training and Inference on AWS

17. May 2026
Google

Foundation model development today scales across three channels: pre-training, post-training, and test-time compute, with AWS showing how its infrastructure—accelerators, networking, and storage—works with open-source tools like PyTorch, Kubernetes, and Prometheus to enable efficient training and inference.

Share on:

Granite Embedding Multilingual R2: Open Apache 2.0 Multilingual Embeddings with 32K Context

17. May 2026
AI Models

IBM releases Granite Embedding Multilingual R2 with two new open-source embedding models (97M and 311M parameters), with the 97M model leading all sub-100M multilingual embedders, both supporting 200+ languages, a 32K-token context window, and code retrieval for 9 programming languages.

Share on:

Asynchronous Batch Processing: Parallel CPU and GPU Utilization for LLM Inference

17. May 2026
AI Models

Asynchronous batch processing enables parallel CPU and GPU work and can increase LLM inference speed by 24 percent without requiring new kernels or models.

Share on:

Anthropic Signs $200 Million Partnership with Gates Foundation

17. May 2026
Anthropic

Anthropic and the Gates Foundation announce a global $200 million partnership to deploy artificial intelligence in health, education, and economic mobility to support millions of people in lower-income countries.

Share on:

PwC Deploys Claude to Develop Technologies, Close Deals, and Redesign Enterprise Functions

17. May 2026
Claude Code

Anthropic and PwC expand their partnership: Claude will be deployed across hundreds of thousands of PwC employees worldwide to develop technologies and transform business operations, with previous implementations showing efficiency gains of up to 70 percent.

Share on:

Claude for Small Business: AI Integration into Proven Business Tools

17. May 2026
Claude AI, Google

Anthropic launches “Claude for Small Business” – an AI solution that integrates into established business tools and supports small businesses with routine tasks such as payroll, invoice management and marketing campaigns.

Share on:

Higher Usage Limits for Claude and Compute Deal with SpaceX

17. May 2026
Claude AI, Claude Code

Anthropic doubles usage limits for Claude Code and increases API rate limits for Claude Opus. A new partnership with SpaceX provides over 300 megawatts of additional compute capacity with more than 220,000 NVIDIA GPUs effective immediately.

Share on:

Ten New AI Agents Automate Financial Services

17. May 2026
AI Models

Anthropic releases ten preconfigured AI agents for financial services that integrate with Microsoft 365 and automate time-consuming tasks. Claude Opus 4.7 leads industry benchmarks and enables implementation in days rather than months.

Share on:

Anthropic Launches New Enterprise AI Company with Blackstone, Hellman & Friedman, and Goldman Sachs

17. May 2026
AI Models

Anthropic is founding a new AI services company with prominent financial partners to help mid-market enterprises integrate Claude into core operations and benefit from advanced AI solutions without internal AI expertise.

Share on:

« Previous
1
…
103
104
105
106
107
…
112
Next »

Adding Benchmaxxer Repellant to the Open ASR Leaderboard

vLLM V0 to V1: Correctness Before Corrections in Reinforcement Learning

EMO: Pretrained Mixture-of-Experts Model for Emergent Modularity

Building Blocks for Foundation Model Training and Inference on AWS

Granite Embedding Multilingual R2: Open Apache 2.0 Multilingual Embeddings with 32K Context

Asynchronous Batch Processing: Parallel CPU and GPU Utilization for LLM Inference

Anthropic Signs $200 Million Partnership with Gates Foundation

PwC Deploys Claude to Develop Technologies, Close Deals, and Redesign Enterprise Functions

Claude for Small Business: AI Integration into Proven Business Tools

Higher Usage Limits for Claude and Compute Deal with SpaceX

Ten New AI Agents Automate Financial Services

Anthropic Launches New Enterprise AI Company with Blackstone, Hellman & Friedman, and Goldman Sachs

Lumi AI News

Legal

Topics