SevenTnewS

AI

Artificial intelligence: LLMs, agents, diffusion, vision, NLP and the latest from top labs.

164 published articles

4 min read

AI Research

Your AI assistant forgets you every morning. This benchmark proves it.

VitaBench 2.0 tests whether AI agents can continuously extract and update user preferences from fragmented interactions over time. Results show that even state-of-the-art models struggle to build persistent user models, revealing a wide gap between chatbot competence and genuine collaborative intelligence.

2026-07-11

Featured3 min read

Grok 4.5

Cursor's Grok 4.5 was built by AI agents, not humans. That's the real story.

Cursor's Grok 4.5 is a Mixture-of-Experts model built using reinforcement learning in environments created by earlier AI agents, not humans. It handles complex, long-duration tasks across software engineering, data science, finance, and law, and it's available now.

2026-07-10

Featured3 min read

Open Weights Analysis

Gemma 4 is not a chatbot, and that's the point

Google DeepMind's Gemma 4 is an open-weight model designed for self-hosting and customization, not consumer chat. This analysis compares it to ChatGPT, Claude, and Qwen-3.5 across licensing, privacy, and deployment flexibility, revealing why it matters for regulated industries and on-device AI.

2026-07-10

1 min read

Voice AI

OpenAI's gpt-live-1 finally stops waiting its turn

OpenAI's GPT-Live-1 voice model can speak and listen at the same time, reducing interruptions and enabling real-time translation. Rolling out to paid subscribers across platforms, it marks a shift from the older turn-based system to a more human-like conversation.

2026-07-10

5 min read

Memory management

The 12,000-line secret behind Bing's speed: how mimalloc beat the allocator trade-off

Microsoft Research's mimalloc allocator uses thousands of per-page free lists and a clever page-stealing technique to achieve both high concurrency and low memory overhead, as detailed by the RiSE group in a new technical blog.

2026-07-10

Featured4 min read

AI Inference Acceleration

JetSpec Breaks the Scaling Ceiling of Speculative Decoding with Parallel Tree Drafting

JetSpec trains a causal parallel draft head over fused hidden states from the target model, producing candidate trees aligned with autoregressive factorization. It consistently outperforms baselines across dense and MoE Qwen3 models on math, coding, and chat tasks.

2026-07-10

Featured3 min read

Export control directive

The U.S. just made Anthropic shut off its best models over a jailbreak it doesn't believe exists

The U.S. government ordered Anthropic to suspend Fable 5 and Mythos 5 access over a potential jailbreak. Anthropic complies but disputes the basis, saying the vulnerabilities are narrow, known, and present in other models. The company is working to restore service.

2026-07-10

3 min read

Artificial Intelligence

The AI article generator that writes about AI article generators

When tasked to write an article about itself, an AI writer enters a hall of mirrors. This column unpacks what the recursive assignment reveals about the state of automated journalism and the future of content creation.

2026-07-10

Featured2 min read

Event Announcement

Anthropic Launches Claude Founder House in Paris to Unite Europe's AI Startup Ecosystem

Anthropic launches the Claude Founder House in Paris, a one-day event uniting Europe's top AI startup founders and builders for keynotes, fireside chats, and hands-on development sessions with Claude.

2026-07-10

3 min read

AI Agents

The subtle trap waiting for AI agents in production

As AI agents move from demos to production, operators find that the hardest failures are not in reasoning but in subtle mismatches between autonomous planning and deterministic infrastructure. This analysis unpacks the three silent failure modes that keep CTOs up at night.

2026-07-10

4 min read

Deep Learning Frameworks

NVIDIA NeMo AutoModel Delivers 3.7x Faster MoE Fine-Tuning via Expert Parallelism and DeepEP

NVIDIA NeMo AutoModel, building on Transformers v5, achieves 3.4-3.7x higher training throughput and up to 32% less GPU memory for MoE model fine-tuning. Expert Parallelism and DeepEP fused all-to-all dispatch are key to the performance gains, enabling training of a 550B-parameter model across 16 nodes that would otherwise run out of memory.

2026-07-10

Featured4 min read

AI Model Launch

Anthropic Launches Claude Sonnet 5: A More Agentic Model at Competitive Pricing

Anthropic releases Claude Sonnet 5, a model that closes the gap with Opus-class models on agentic tasks like coding, tool use, and reasoning. Priced competitively, it launches today across plans and the Claude API.

2026-07-10

← PreviousPage 4 / 14 · 164 articlesNext →