-
MiniMax M3 is an open-weight model that beats GPT-5.5 on SWE-Bench Pro with a 1M-token context
MiniMax M3, out June 1 from the Shanghai lab, makes a loud claim: the first open-weight model to put frontier coding, a 1-million-token context window, and native multimodal understanding in one architecture. On SWE-Bench Pro it scores 59.0%, edging out OpenAI’s GPT-5.5 and Google’s Gemini 3.1 Pro. ## The benchmark sweep It’s not just SWE-Bench.… Continue reading
-
ArcANE tests whether role-playing AI characters evolve with the story, not just stay in character
Role-playing AI is usually judged on consistency: does it remember the character’s facts and stay on persona? ArcANE, a new arXiv benchmark, argues that’s the wrong bar. A good character isn’t fixed — its values and behavior should evolve as the story moves, and the real question is whether the AI tracks that arc at… Continue reading
-
VideoKR builds the first large training corpus for knowledge- and reasoning-intensive video understanding
Most video AI gets graded on shallow recall — what’s on screen at minute three. VideoKR, a new arXiv release, targets the harder thing: video questions that need outside knowledge and multi-step reasoning, not a textual shortcut. It’s billed as the first large-scale training corpus built specifically for that. ## What’s in it The dataset… Continue reading
-
Brandfetch’s Brand Context API gives AI agents structured data on 50M brands in one call
LLMs are confidently wrong about brands they barely know — the logo, the positioning, the competitors. Brandfetch’s Brand Context API is built to ground that: one API call returns structured data on 50M+ brands, including voice, mission, positioning, audience, and competitors. ## What it’s for The pitch is grounding AI features so they stop hallucinating… Continue reading
-
Universal Memory Protocol wants to be MCP for agent memory
AI agents still can’t carry memory across sessions, agents, or vendors — every harness reinvents it privately. The Universal Memory Protocol (UMP) is a bid to fix that the way MCP standardized tools: a transport-neutral protocol for how agents store and recall memory. ## What it standardizes UMP defines negotiated memory operations over a portable,… Continue reading
-
Code2LoRA generates a repo-specific adapter so code models keep up as the codebase changes
Code models need to know your repository — its imports, APIs, and conventions — to be useful. The usual fixes are clumsy: stuff all that context into a long prompt (RAG, dependency crawls), or fine-tune a LoRA per repo, which is expensive and goes stale the moment the code changes. Code2LoRA, a new arXiv paper,… Continue reading
-
Manus’s Shopify connector builds and runs a whole store from one chat
Manus added a Shopify connector that turns running an online store into a conversation. Tell it what you sell, who it’s for, and the vibe you want, and by the end of the chat you have a full draft store — branding, products, pages, a working cart — sitting in your own Shopify account. ##… Continue reading
-
MLEvolve is a self-evolving agent that hit #1 on MLE-bench in 12 hours
MLEvolve is an open-source multi-agent system that designs machine-learning solutions end to end — planning, coding, validating, and iterating the way a human ML engineer would on a Kaggle problem. Built by Shanghai AI Laboratory and East China Normal University, it reached #1 on the MLE-bench leaderboard with 12 hours of runtime. ## How it… Continue reading
-
Grok Imagine 1.5 turns a still image into 720p video with synced audio and tops the image-to-video arena
xAI’s Grok Imagine 1.5 takes a still image (or a text prompt) and animates it into a clip — with native, synchronized audio baked in: music, sound effects, even lip-synced dialogue. It shipped as an API preview on June 3. ## What’s new in 1.5 The headline is audio in every generation — no separate… Continue reading
-
Extella chains specialized AI agents into one pipeline with memory that compounds
Extella is an AI execution platform built around one idea: connect specialized agents into a single pipeline and let the system get smarter the more you use it. No glue code, no tool sprawl — words in, results out. ## The dynamic system prompt Its standout claim is a “dynamic system prompt.” Instead of static… Continue reading
Top AI Product
Every day, hundreds of new AI tools launch across Product Hunt, Hacker News, and GitHub. We dig through the noise so you don't have to — surfacing only the ones worth your attention with honest, no-fluff reviews. Explore our latest picks, deep dives, and curated collections to find your next favorite AI tool.
Top AI Product 