Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
A new system architecture called WhatsApp Coding enables mobile-first development by using a secure proxy chain between WhatsApp, a GCP VM, and a local MacBook, allowing technical executives to validate autonomous multi-…
Researchers at MIT found that large language models develop a modular cognitive architecture mirroring the human brain, with distinct neuron sets supporting language, formal reasoning, social reasoning, and physical reas…
Anthropic restored its Fable 5 AI model globally on July 1 after the US government lifted export controls that had suspended access on June 12. The controls were triggered by a reported jailbreak, which Anthropic later f…
Cloudflare's Durable Objects now stay alive for the full duration of active outbound connections, eliminating the need for heartbeat code to prevent eviction during LLM token streaming. Zhipu's GLM-5.2 model achieves 39%…
OpenAI is expanding its advertising business beyond a single format, with job listings indicating plans for text, image, video, native, conversational, and interactive ads. The company is hiring software engineers to bui…
Hong Kong's tech chief Sun Dong warned that artificial intelligence will trigger the 'greatest industrial revolution,' surpassing all previous ones, and will significantly impact graduates and workers in traditional sect…
Researchers developed a closed-form model of Group Relative Policy Optimization (GRPO) training dynamics, predicting reward trajectories and stability thresholds from first principles. The model subsumes empirical satura…
Researchers introduced TheraJudge and TheraAgent, a framework that uses multi-agent systems and human-aligned evaluation to improve therapeutic quality in mental health support LLMs, achieving a +0.43 improvement in clin…
Researchers introduced RoPoLL, a robust panel of LLM judges that replaces standard consensus aggregation with a geometric median estimator, achieving unbounded bias resistance under contamination. In tests across 13 open…
Researchers introduced AgRefactor, an LLM-based multi-agent workflow that refactors software into HLS-compatible code with a self-evolving memory system. On 9 of 11 challenging benchmarks, it outperformed state-of-the-ar…
Researchers released a benchmark for Arabic-Russian scientific translation, including a 27,000-sentence parallel corpus and fine-tuned multilingual models. The Qwen2.5-7B model achieved BLEU 23.15, outperforming zero-sho…
Researchers introduced Contrastive Reflection, an iterative prompt-optimization framework for agentic information retrieval workflows, which uses error-anchored behavioral slices and contrastive examples to propose targe…
Researchers introduced OpenLife, a proof-of-concept system that uses LLM agents with persistent memory, tool use, and a budget-based metabolism to enable open-world artificial life. Running six agents for twelve weeks, t…
Researchers introduced a Bangla event detection benchmark with 9,979 annotated sentences across clean text, ASR transcripts, and corrupted text. They found encoder models perform better on clean text but degrade under no…
Researchers introduced Indi-RomCoM, a benchmark for evaluating large language models on Romanized code-mixed Indic-English instructions. Testing across seven tasks and four languages showed that LLMs consistently underpe…
Researchers introduced GradeSQL, a framework for training Outcome Reward Models (ORMs) to improve test-time verification in Text-to-SQL tasks. ORM-based selection outperformed execution-based Best-of-N and Majority Votin…
Researchers deployed an automated pipeline to optimize skill descriptions for an enterprise AI agent, achieving 79.2% F1 accuracy versus 79.4% for manual tuning while reducing engineering effort per skill from 120 to 3.8…
Researchers at arXiv trained GPT-2 models on 'impossible' languages and found that while grammatical sensitivity degrades gradually, generative production fails significantly, suggesting transmission failures explain why…
Researchers submitted a system to SemEval-2026 Task 9 for detecting multilingual online polarization using transformer-based models with class weighting and threshold tuning. Their approach achieved competitive F1 macro …
Researchers extend Frictive Policy Optimization (FPO) to handle perceptual asymmetry in dialogue, where participants have asymmetric partial information. Cross-corpora analysis and LLM probing on the HCRC MapTask show th…