AI news for builders and product teamsUpdated Nov 18, 2026, 10:01 UTC
AI news, straight from the source.
Releases, research, and ideas for developers and product teams. Updated every hour.
Across the feeds
Newest first
A new study by MIT and other researchers found that AI assistance improved skin disease diagnosis accuracy for non-experts and clinicians, but explainability methods had different impacts depending on user knowledge level. Non-experts deferred to AI, especially LLM explanations, while clinicians performed best with only a model prediction and no explanation.

Interconnects has launched two free data projects: an Artifacts Hub covering 792 open models released in the last two years, and a daily-updating Adoption Dashboard tracking downloads and derivative models by geography and organization.

Episode 253 of the Last Week in AI podcast, recorded 07/29/2026, covers major AI releases including Anthropic's Claude Opus 5 and Google's Gemini 3.6/3.5 Flash variants, plus compute deals, open-weight models and policy and safety news.

A roundup of recent open model releases covers Thinking Machines' Inkling, Poolside's Laguna S2.1 and Laguna-XS-2.1, Tencent's Hy3, DeepSeek-V4-Flash-0731, Kimi K3, Meituan's LongCat-2.0, and Motif-3-Beta, among others.

Moonshot AI's Kimi K3, described as a 2.8-trillion-parameter open-weights model and the first open-source model in the 3-trillion-parameter class, is now served on Together AI via an OpenAI-compatible API. The guide covers KDA and Attention Residuals architecture, Stable LatentMoE sparsity, reasoning effort levels, 1M context, tools, vision, benchmarks, and per-token pricing.

Together AI added autoscaling to its Dedicated Model Inference endpoints, letting deployments scale on inference-native metrics like in-flight requests and TTFT, with configurable replica bounds and scale-up/down windows. A replay experiment showed that only the in-flight-requests policy scaled under bursty load, while TTFT and GPU-utilization policies stayed flat.

Amazon researchers presented ControlG at ICML, a framework that uses PID controllers from industrial control systems to schedule multiple training objectives in graph self-supervised learning by allocating compute to one objective at a time instead of blending gradients. It outperformed baselines on nine graph benchmarks across node classification, link prediction, and clustering.

Amazon Science introduced PatientAgentBench, a clinician-vetted benchmark for evaluating patient-facing health AI agents across multiturn conversations, and reported that even capable frontier models fell short of required safety standards.

Together AI and academic collaborators introduced ThunderAgent, a program-aware scheduler for agentic inference that treats each agent workflow as a schedulable program. It reports over 2x single-node throughput, 2.4x speedup on an 8-node cluster, and acceptance as an ICML 2026 Spotlight paper.

Together AI announced a strategic partnership with Moonshot AI to natively serve Kimi models, beginning with Kimi K3, a 2.8T parameter sparse Mixture-of-Experts model with native vision and a 1M token context window. Together AI becomes a launch platform for Moonshot's open weights releases, offering day zero access and post-training.

Epoch and METR released MirrorCode, a benchmark testing whether AI systems can reimplement programs from CLI access alone; some tasks were solved, but 8 of 25 targets were never fully solved. Anthropic, Sunday, and OpenAI also reported robotics and safety findings.

Amazon is providing long-term financial support to the Lean Focused Research Organization to advance the Lean programming language for mathematical proofs of software correctness. Amazon says the donation is the largest in the FRO's history and that Lean-based verification is used in Bedrock AgentCore, SampCert, and AWS Neuron.

Ethan Mollick's Summer 2026 guide says the practical frontier for getting real work done with AI is now agentic systems — Claude's Cowork/Code and ChatGPT's Work/Codex — where an AI model is given a computer and tools to plan and act.

Together AI released an update to its inference platform for running open-weight models in production, with rollout controls, autoscaling, and observability. It also opened a closed beta for custom training, including full-weight and LoRA reinforcement learning and supervised fine-tuning.

An Interconnects podcast episode recaps open-model developments, covering Kimi K3's release and usage, GLM 5.2, Chinese labs including Qwen and DeepSeek, the open-closed performance gap, distillation debates, and cybersecurity arguments against model bans.

Last Week in AI's episode 252, recorded 07/11/2026 and hosted by Andrey Kurenkov and Jeremie Harris, covered OpenAI's GPT-5.6 rollout and ChatGPT Work rebrand, SpaceX AI's Grok 4.5, Meta's Muse releases, and AI policy developments including AI 2040.

Episode 251 of the Last Week in AI podcast covers Anthropic redeploying Claude Fable 5 and launching Claude Sonnet 5, new Google tools, and business and research updates including Etched, Baidu, Agility Robotics, DeepSeek and China's LongCat-2.0.

Last Week in AI released its 250th episode, recorded 06/27/2026 and hosted by Andrey Kurenkov and Jeremie Harris. Topics included US government gating of frontier AI, OpenAI's GPT-5.6 Sol rollout, Anthropic's Mythos, compute supply chain competition, and GLM 5.2.

Episode 249 of the Last Week in AI podcast, recorded 06/17/2026, covers Anthropic cutting off access to Fable 5 and Mythos 5 after a US government order tied to alleged jailbreaks, and SpaceX's IPO at a roughly $1.75T valuation followed by a move to acquire Cursor for $60B.

Last Week in AI released episode 248, recorded 06/12/2026, covering Anthropic's Claude Fable 5 release, Apple's Siri AI announcement at WWDC, and a range of business, open-source, policy and safety news items.