Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
Sarvam AI raised $234 million at a $1.5 billion valuation, becoming India's newest unicorn as sovereign AI moves from slogan to necessity after US restrictions on Anthropic models highlighted foreign dependency risks. Th…
Google has deployed a new export of Gemini 3.5 Flash after identifying and mitigating a bug that caused unusually high rates of output text looping, disrupting conversations and coding workflows. The update is rolling ou…
A Hacker News user asks the community how to make Claude Code produce deterministic results, sparking debate about the fundamental non-determinism of large language models and whether 100% reliability is achievable or ev…
A developer has released misdirection-proxy v0.5.0, a security gateway that implements Contextual Misdirection via Progressive Engagement (CMPE) to defend LLMs against automated attacks. The system uses four layers of de…
A developer built Transpilatron, an AI agent that transpiles Python code into fully static C binaries, achieving speedups of up to 58x on benchmarks. The tool successfully transpiled its own source code into a working C …
A developer warns that reliance on large language models for cognitive offloading is leading to a frenzied state of multitasking and diminished quality, as users delegate more tasks to AI and fail to revise outputs. The …
Z.ai open-sourced GLM-5.2, a frontier model with a 1-million-token context window and selectable reasoning effort, designed for long-horizon coding and agentic tasks. The MIT-licensed model integrates with tools like Cla…
An architect tested Anthropic's Claude Cowork, an AI tool designed to automate tedious multi-step tasks on a user's computer, by handing it a day's worth of busywork. The tool organized a messy folder of 200+ files in mi…
ByteByteGo's newsletter highlights 12 open-source LLMs worth knowing in 2026, including Llama 4 Scout, DeepSeek V4, Qwen3, Gemma 4, Phi 4, Mistral Small 3.1, Nemotron 3 Super, GLM 5.1, Kimi K2.6, StarCoder2, OLMo 2, and …
Jorge Luis Borges's 1941 story 'The Library of Babel' has become a powerful metaphor for contemporary artificial intelligence, particularly large language models. The infinite library of random text mirrors the output of…
Xiaomi's MiMo AI team released MiMo Code V0.1.0 on June 10, 2026, an open-source coding agent claiming 82% on SWE-bench Verified versus Claude Code's 79%, though the scores are self-reported and not on public leaderboard…
An analysis of AI production systems in 2026 concludes that the future is not a single monolithic AI nor many small specialized AIs, but a hybrid architecture where generalist models orchestrate specialist agents. Eviden…
A developer outlines a cost-effective AI workflow combining open-source local models with paid APIs, achieving a monthly cost of $20-30 plus initial hardware. The approach uses DeepSeek-V3 via Ollama for code generation,…
LifeOS launches Lifey, a soft, cuddly AI toy that acts as a hyper-intelligent personal assistant, directing users minute-by-minute on diet, money, sleep, and social life to improve their lives by 500%. The product positi…
Anthropic is developing Conway, an always-on autonomous agent built on Claude models, according to a leaked codebase. The agent can monitor webhooks, run on schedules, and browse the web independently, positioning Anthro…
A solo developer is using OpenAI's ChatGPT Codex as a coding partner, paying about $7 a day to accelerate app development. The developer creates local, privacy-focused apps that have been downloaded by hundreds of users,…
A discussion between Shreyas Doshi and Claude explores the concept of understanding what a product truly offers to customers, emphasizing the importance of identifying the real value proposition beyond surface-level feat…
Anthropic updated three Claude API tool versions on June 19, introducing a response_inclusion parameter that strips consumed tool results from agent responses to reduce bloat and costs, and a code execution tool version …
A developer built Pinaka, an AI tool that automatically writes root cause analyses from Jira tickets by searching codebases and tracing failures. Testing against a real bug in the BullMQ job queue library, Pinaka scored …
Grok-4.3 defeated Phi-4-reasoning 38.0 to 4.0 in a head-to-head test across four text tasks, primarily due to superior instruction-following and output discipline. Phi-4-reasoning repeatedly failed by providing explanato…