Greetings Qrew,

Whether you're diving into AI for work, side projects, or just out of curiosity — this issue of The Creators Hive brings you practical insights, hands-on learning, and tools you can actually use.

1. OpenAI to allow erotica on ChatGPT
2. Ant Group’s thinking model bags IMO silver
3. Google’s upgraded Veo 3.1 video model
4. Anthropic’s fast, low-cost Claude Haiku 4.5
5. Google’s Gemma-based AI finds new cancer treatment
6. Tool Recommendation of the Day: Anime.js

---

OpenAI to allow erotica on ChatGPT

OpenAI CEO Sam Altman announced upcoming updates to ChatGPT’s personality system, including a new feature that will permit mature and erotic conversations with verified adult users — a shift reminiscent of xAI’s Grok companions.

Key details:

• Altman acknowledged that ChatGPT’s past restrictions were meant to protect at-risk users but said they also limited enjoyment and authenticity for adults.
• OpenAI plans to roll out a more expressive version of ChatGPT first, inspired by the user-favorite 4o personality, followed by mature mode support by December.
• The erotica feature will be available only to verified adults with built-in age checks and safety mechanisms.
• This follows OpenAI’s earlier statement allowing developers to build “mature” GPTs and apps inside ChatGPT, provided they meet safety and consent standards.

Why it matters:​
The move positions OpenAI squarely in the booming AI companion market, where emotional and intimate interaction drives engagement. While Altman insists the mode will activate only by explicit user request, the sheer scale of ChatGPT’s 800M weekly users raises serious questions about safeguards, consent, and the long-term psychological effects of AI-driven intimacy.

---

Ant Group’s thinking model bags IMO silver

Ant Group’s AI division InclusionAI has introduced Ring-1T, an open-source “thinking model” that reached silver-level performance on the International Mathematical Olympiad (IMO) benchmark — placing it just behind Google and OpenAI’s gold-tier systems.

Key details:

• Built on InclusionAI’s Mixture of Experts (MoE) design, Ring-1T features 1 trillion parameters (with 50 billion active per task) and a 128K context window for extended reasoning and multi-turn problem solving.
• On the IMO test, it solved four problems on the first attempt and one after three tries, achieving silver standing and nearly matching Gemini 2.5 Pro on the toughest questions.
• The model also posted near–state-of-the-art (SOTA) results across multiple reasoning and logic benchmarks, outperforming or rivaling several leading systems from Anthropic, Google, and OpenAI.
• InclusionAI confirmed that Ring-1T is still in active training, with ongoing improvements focused on alignment, reasoning depth, and energy efficiency.

Why it matters:​
Ring-1T’s performance narrows the gap between open-source and frontier closed models, signaling a new level of parity in global AI research. It’s also a major milestone for China’s AI ambitions, positioning Ant Group’s InclusionAI as a credible challenger in high-level reasoning — and edging the world closer to an internationally balanced AI ecosystem.

---

Google’s upgraded Veo 3.1 video model

Google has launched Veo 3.1, the newest version of its video generation model, delivering enhanced visual realism, smoother motion, and more precise creative control — especially for filmmakers and content creators.

Key details:

• Character consistency: Veo 3.1 can now take up to three reference images to ensure characters remain consistent across multiple scenes.
• Start and end frame transitions: Users can supply both beginning and ending frames, allowing the model to generate seamless transitions — complete with synchronized audio.
• Scene extension: The model supports continuous scene expansion, enabling the creation of videos up to one minute long by stitching together multiple clips.
• Accessibility: Both standard and fast versions of Veo 3.1 are being rolled out through Google’s Flow filmmaking tool, Vertex AI, and Gemini.

Why it matters:​
While Sora 2 recently dominated headlines for its cinematic realism, Veo 3.1 focuses on practical creative control — letting users refine, extend, and direct scenes with precision. The model’s emphasis on editing and continuity over raw spectacle hints at Google’s strategy to position Veo as the preferred tool for professional storytelling, not just viral visuals.

---

Anthropic’s fast, low-cost Claude Haiku 4.5

Anthropic has launched Claude Haiku 4.5, the smallest member of its latest model lineup — offering flagship-level performance at a fraction of the cost and with dramatically faster speeds, signaling another big step in AI efficiency.

Key details:

• Performance parity: Haiku 4.5 matches the coding and reasoning abilities of Claude Sonnet 4 (released in May) while costing just $1 per million input tokens, compared to Sonnet’s $3.
• Benchmark gains: Despite being smaller, Haiku 4.5 outperforms Sonnet 4 on tasks involving computer use, math, and agentic tool execution, and even approaches GPT-5-level scores in select evaluations.
• Agentic orchestration: Enterprises can deploy multiple Haiku agents in parallel, coordinated by Sonnet 4.5, to handle multi-step or large-scale workflows efficiently.
• Availability: Haiku 4.5 is now accessible to all Claude users — including the free tier, as well as through Claude Code and API integrations.

Why it matters:​
Claude Haiku 4.5 embodies the rapid deflation of intelligence costs in AI — where models once deemed “premium” are now cheap, fast, and widely accessible. Anthropic’s focus on scalable, low-cost intelligence could reshape enterprise economics and accelerate the shift toward agentic automation at scale.

---

Google’s Gemma-based AI finds new cancer treatment

Researchers from Google and Yale University unveiled C2S-Scale 27B, a Gemma-based foundation model that uncovered a previously unknown cancer treatment pathway — one that has already been validated in living cells.

Key details:

• Biological language modeling: C2S reads cellular data like a language, modeling how cells behave, communicate, and respond to drugs.
• Research goal: Scientists asked the model to find compounds that help the immune system better recognize tumors, but only under specific signaling conditions.
• Breakthrough: The model identified silmitasertib, a known drug never before associated with immune-based cancer detection.
• Validation: Experiments confirmed the prediction — the drug combo made tumor cells ~50% more visible to immune defenses, a major potential advance for cancer immunotherapy.

Why it matters:​
This marks a milestone moment for AI-driven science — moving from analyzing existing data to generating novel biological discoveries. With Google noting that biological foundation models scale predictably like language models, the next frontier could be AI as an active collaborator in medical research, not just a computational assistant.

---

Tool Recommendation of the Day: Anime.js

Website: animejs.com

What makes it unique

Anime.js is a lightweight, high-performance JavaScript animation library that works with CSS properties, SVG, DOM attributes, and JavaScript Objects. Its intuitive API simplifies creating sophisticated animations such as motion paths, shape morphing, staggered sequences, and scroll-triggered effects.

Key Features

• Animate almost anything: CSS transforms, SVG drawings, JavaScript values.
• Built-in timeline support, stagger utilities, and easing options.
• Modular imports allow you to include only what you need, keeping bundle size small.
• Compatible with modern workflows (npm, bundlers) and simple script tags.
• Supports advanced techniques: drag & drop, springs, scroll-based triggers, responsive animations.

Use Case Magic

• Enhance websites with rich animations: hero sections, interactive elements, micro-interactions.
• Create animated infographics or engaging data visuals on the web.
• Prototype motion design without needing full animation software.
• Educators or developers can use it to animate UI transitions, tutorials, or interactive demos.

Why it stands out

While many animation libraries focus on either simple CSS transitions or heavy frameworks, Anime.js strikes a balance: it’s both powerful and accessible. The modular design ensures you can scale from simple effects to complex sequences without significant overhead.

---

Masterpieces from the Hive 🎨

Here’s what our community has been creating this week using the Hive Wall:

32 North Gould Street, Sheridan, Wyoming 82801
​Unsubscribe · Preferences