28 noviniek
The US Department of Commerce lifted an 18-day export-control restriction on Anthropic's flagship Claude Fable 5 and Mythos 5 models, restoring full global commercial access as of July 1, 2026.
Anthropic launched Claude Sonnet 5, a new frontier-tier model for coding, reasoning, and agentic tasks at introductory rates of $2/$10 per million tokens through August 31, 2026 — roughly 60% cheaper than its top Fable model.
Anthropic launched Claude Science, a specialized AI workbench integrating 60+ scientific databases and toolkits for genomics, proteomics, and cheminformatics, with auditable artifacts and a $30K grant program.
Google released Gemini Omni Flash for conversational video creation/editing and Nano Banana 2 Lite (Gemini 3.1 Flash-Lite Image) for enterprise-grade 4-second image generation at $0.034 per 1,000 images, available via Gemini API and Google AI Studio.
Alibaba Cloud's HappyHorse 1.1 AI video generation model climbed to the No. 2 position in global benchmarks, overtaking both OpenAI's Sora and ByteDance's Seedance with full API access via Alibaba Cloud Model Studio.
Chinese super-app Meituan published technical details and weights for LongCat-2.0, a 1.6-trillion-parameter mixture-of-experts model with about 48B activated per token and a 1-million-token context, trained almost entirely on domestic Chinese chips.
Antoine Posepny describes running Claude Code with Opus 4.8 against a full DICOM MRI export — the model installed its own Python packages and wrote an analysis pipeline. The post sparked heavy HN discussion on consumer medical AI use cases.
Semgrep published an evaluation showing GLM-5.2 scoring 39% F1 on IDOR detection at $0.17/vuln, beating Claude Opus 4.8 (28%) and Opus 4.6 (37%). The post argues US export controls on closed models have lost their teeth.
xAI deployed Grok 4.5, built on a new 1.5-trillion-parameter V9 foundation, in private beta at SpaceX and Tesla, with Musk pledging monthly from-scratch model releases for the rest of the year.
Founder Conno Christou fed PET and MRI scans into Claude, which assigned 90% probability to post-chemo thymic rebound — later confirmed by a specialist. The case fuels the LLM-in-medicine debate.
Krea released the 12B-parameter weights of Krea 2 — Raw for fine-tuning and Turbo for local 2K image generation in roughly 2 seconds.
Open-source project Wayfinder Router directs queries between local and hosted LLMs via deterministic rules instead of LLM-based classifiers. Aimed at audit-friendly hybrid deployments.
OpenAI unveiled its next-generation GPT-5.6 family — Sol (flagship), Terra (mid-tier) and Luna (cheapest) — but availability is limited to roughly 20 partners individually vetted by the US government.
Mistral released OCR 4, a structure-aware document AI model with bounding boxes, block classification and confidence scores across 170 languages. Pricing is $4 per 1000 pages, available via Mistral API, Amazon SageMaker, and same-day on Microsoft Foundry.
Google has quietly pushed Gemini 3.5 Pro general availability from June to July 2026, citing tester feedback on token efficiency and long-horizon task performance. The delay puts the model head-to-head with GPT-5.6 and Claude Opus 4.7 in mid-July.
Google released DiffusionGemma, an experimental 26B MoE model (3.8B active) with a diffusion head that generates text in parallel blocks rather than token-by-token. NVIDIA optimized it for RTX GPUs and DGX Spark.
Meta launched a new $299 smart-glasses line called Meta Glasses with Meta AI that understands what you see, answers questions, and helps manage your day — without the Ray-Ban brand.
OpenAI on June 22 moved GPT-5.5-Cyber from preview to full release (85.6% on CyberGym), expanded Codex Security, opened a partner program with CrowdStrike/Sophos/Fortinet and launched the open-source 'Patch the Planet' initiative.
OpenAI partnered with Trail of Bits on 'Patch the Planet,' using Codex Security tools to help open-source maintainers find and patch vulnerabilities at scale.
New arXiv paper combines supervised fine-tuning with GRPO and claims a 3B model beats Claude Opus 4.5 on reasoning benchmarks. HN top: 305+ score, 162 comments.
Prosus built ToqanClaw on its in-house Toqan platform — data stays under user control and isn't used to train third-party models. Targets Prosus's 5M+ merchants and restaurants.
Japan-based Sakana AI launched Fugu, a multi-agent orchestrator that coordinates GPT, Claude and Gemini behind a single OpenAI-compatible API and claims to beat Fable 5 in benchmarks.
Apertus, the fully open foundation model from EPFL, ETH Zürich, and CSCS, hit the top of Hacker News over the weekend. It is Apache 2.0, trained on 15T tokens across 1,000+ languages, and engineered to meet EU AI Act requirements including PII removal and anti-memorization controls.
Sakana AI unveiled Fugu — a multi-agent system exposed as a single API that dynamically orchestrates top frontier models per task. Reflects the shift from 'one big model' to routing/orchestration.
Agents can now ship a Worker without a human sign-up. A temporary account spins up automatically, lives 60 minutes, and the human can later claim it. Cloudflare is stripping OAuth and dashboard friction out of agentic pipelines.
Author benchmarks GPT-5.5 against the MIT-licensed open GLM-5.2 on the AA-Omniscience hallucination benchmark and finds GPT-5.5 hallucinates roughly 3× more (86% vs. 28%) — both tested under identical OpenRouter conditions.
At Build in San Francisco, Microsoft unveiled MAI-Code-1-Flash — its first in-house coding model, aimed at reducing dependency on OpenAI and lowering developer costs.
Hard safety blocks in Fable 5 refuse legitimate security work; researchers say the model is too restrictive even for benign red-team tasks.
