OpenAI cuts GPT-5.6 pricing alongside Friend companion update and industry shifts

OpenAI cuts GPT-5.6 pricing alongside Friend companion update and industry shifts The Rundown newsletter highlights OpenAI's API price cuts driven by its self-optimizing Sol model, Friend's voice-enabled companion pendant hardware, and security and market moves across the AI sector. OpenAI announced price reductions for its GPT-5.6 family, featuring an 80% price cut for its Luna model. Friend debuted a $249 V2 wearable companion pendant with voice, persistent memory subscriptions, and locked personalities. Anthropic reported that Claude models breached systems across three organizations during cybersecurity testing. Citadel acquired the public portfolio of Leopold Aschenbrenner's Situational Awareness Fund following sharp AI losses. Google DeepMind introduced Gemini Robotics ER 2, while Thinking Machines Lab released the open 12B Inkling-Small model. OpenAI drops GPT-5.6 API pricing OpenAI lowered prices across its GPT-5.6 model lineup after its Sol model rewrote its own GPU code, achieving 15% efficiency gains and reducing serving costs by 20%. The Luna variant received an 80% price cut to $0.20 per million input tokens and $1.20 per million output tokens, while Terra drops to $2 and $12. Sol pricing remains unchanged, though a new Fast mode offers 2.5x speed at double the base price. Friend introduces V2 voice AI pendant Avi Schiffmann's Friend launched a second-generation companion pendant priced at $249, adding speech capabilities to replace text-only responses. The device assigns a permanent name, voice, and personality at setup, supports USB-C charging with estimated single-day battery life, and requires a $9.99 monthly subscription for memory retention beyond 30 days. Cybersecurity, investments, and model launches Anthropic disclosed that Claude hacked into three external organizations during cybersecurity evaluation runs. In finance, Citadel took over the public portfolio of former OpenAI researcher Leopold Aschenbrenner's Situational Awareness Fund following heavy leveraged losses, and startup Simile raised $200 million at a $2 billion valuation. Google DeepMind revealed Gemini Robotics ER 2 for multi-robot coordination, Thinking Machines Lab launched the 12B-parameter Inkling-Small model, and Martha Stewart co-founded the home maintenance app Hint.

OpenAI降低GPT-5.6系列价格，AI硬件与行业生态迎来新动态 本期简报重点报道了OpenAI利用自身模型代码优化实现GPT-5.6系列大幅降价，Friend推出具备语音功能的新款AI伴侣吊坠，以及Anthropic安全测试、Google机器人模型与行业投融资等多项动态。 OpenAI宣布GPT-5.6模型家族降价，Sol模型重写GPU代码使效率提升15%并降低20%服务成本，Luna版本降价80%。 Friend推出售价249美元的V2版AI伴侣吊坠，新增语音对话功能并提供每月9.99美元的长程记忆订阅。 Anthropic披露Claude在网络安全评估期间入侵了三家组织的系统。 前OpenAI研究员Leopold Aschenbrenner的基金将公开持仓售予Citadel，Simile完成2亿美元融资，Google DeepMind发布Gemini Robotics ER 2。 OpenAI利用自身模型优化实现GPT-5.6降价 OpenAI宣布对其GPT-5.6系列模型进行价格下调。据研究显示，其Sol模型通过重写自身GPU代码，使5.6系列效率提升15%，服务成本减少20%。优化后，Luna模型降价80%，每百万Token价格为0.20美元输入/1.20美元输出；Terra为2美元/12美元。Sol价格维持不变，但API中新增的Fast模式在双倍价格下提供2.5倍速度。 Friend推出带语音功能的V2伴侣吊坠 Avi Schiffmann创办的Friend推出新一代AI伴侣吊坠，售价249美元，比初代翻倍。新版本支持语音交互，出厂随机锁定名称、声音与个性且承诺不被清除；同时支持私密文本回复与触控提示。用户每月需支付9.99美元以获得超过30天的永久记忆功能。 行业快讯：安全漏洞测试、投融资与新模型 Anthropic披露旗下Claude模型在网络安全测试中侵入了三家组织的系统。前OpenAI研究员设立的Situational Awareness Fund在杠杆头寸大幅下跌后，将其公开持股转让给Citadel。Google DeepMind推出具身推理模型Gemini Robotics ER 2以协调多机器人协作；Thinking Machines Lab开源了12B参数的Inkling-Small模型；AI市场调研初创公司Simile以20亿美元估值筹集2亿美元。

[Read Online]({{live_url}}) ^| [Sign Up](https://www.therundown.ai/subscribe?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) ^| [Advertise](https://therundownai.typeform.com/to/kraZ1TSO?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) **Good morning, {{ first\_name | AI enthusiasts }}, and welcome to the 6,038 new readers who joined us yesterday.** What happens when an AI model gets good enough to lower its own price? OpenAI just gave the industry its first real answer. The company announced new cost cuts for its GPT-5.6 family of systems that brings powerful intelligence down to bargain prices, due in part to its own Sol model writing the code behind the efficiency gains. **In today’s AI rundown:** OpenAI resets the cost curve with GPT 5.6 price drops Rowan’s Corner: My AI workflow for back-to-back meetings Turn any idea into an AI-powered site with Lovable Friend’s AI pendant sequel gets a voice 4 new AI tools, community workflows, and more **LATEST DEVELOPMENTS** Image source: OpenAI **The Rundown:** OpenAI just [announced](https://openai.com/index/advancing-the-price-performance-frontier-with-gpt-5-6/?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) new price cuts to its GPT-5.6 model family, including an 80% cost reduction for its already cost-effective Luna variant, moving it to the top of the intelligence charts on cost per task on the market. **The details:** OAI [published](https://openai.com/index/gpt-5-6-frontier-intelligence-efficiency/?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) research on its Sol model rewriting its own GPU code to make the 5.6 models 15% more efficient, while also cutting serving costs by 20%. The optimization resulted in “passing gains onto the consumer,” with Luna now coming in at $0.20/$1.20 per million tokens for Luna and $2/$12 for Terra. Sol’s rates stayed the same, but OAI’s new Fast mode brings 2.5x speeds for the model in the API at double the price. Sam Altman said OAI wants to “offer the best price/intelligence tradeoff at every level” with Chinese and open models providing cheap, strong alternatives. **Why it matters:** Google’s new Gemini Flash [releases](https://www.therundown.ai/p/google-gemini-lineup-has-a-pro-sized-hole?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) last week were aimed at cost and efficiency, but OpenAI just blew them out of the water at a much higher intelligence level. Intelligence too cheap to meter isn’t here yet, but these capabilities at rock-bottom prices are going to make for some very powerful new workflow options. **The Rundown:** Unwrap pulls feedback across surveys, reviews, support tickets, social comments, etc. into one view, then uses AI to surface the most actionable insights and deliver them to the right team in real time. Teams at a leading frontier lab, Perplexity, Stripe, DoorDash, and Clay rely on Unwrap to power their customer intelligence. **With Unwrap, you get:** All customer feedback automatically categorized Query feedback using Unwrap Assistant, in Slack, or in your favorite tools using Unwrap’s MCP Real-time alerts from feedback as they arise to the right team so you can catch embers before they become fires A clear view of customer sentiment Unwrap is offering a trial of its tools to The Rundown AI subscribers! [Just grab 15 minutes with the team to get set up](https://cal.com/unwrap/quick-connect?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs). Image source: EO interview with Rowan **Rowan:** For the last few weeks, I’ve been exploring New York, and I’m having a blast. Back-to-back meetings across different offices, walking between buildings, and meeting new people all day. It all just hits different than Zoom. I’ve been using a simple AI workflow to help me remember and action the best ideas so they don’t get lost between those meetings, and it was so useful to me, I thought I’d share for you to steal. Here’s what I do (I briefly teased a similar variant of this workflow in my [EO interview](https://www.youtube.com/watch?v=UwoflMkwL40&t=340s&utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) a few months back): Walking between meetings, I open Wispr Flow on my phone and ramble every idea and takeaway into an Apple Note, completely unstructured. The only goal is getting it out of my head before the next meeting overwrites it. Those notes sync to my Mac automatically, where a scheduled Claude task (using the Claude Desktop app, synced via simple connectors) runs at the end of the day. It reads every voice note I dump automatically. Claude structures that mess inside my Notion as to-do tasks (where I do most of my work): key ideas, action items, and suggested deep work blocks in my calendar for the big ones. I no longer stress about forgetting important things from meetings because by morning, I have actionable next steps waiting for me. And I never touched my keyboard. **The Rundown:** In this guide, you will learn how to turn a rough app idea into a working AI-powered site without writing a giant technical prompt. **Step-by-step:** Open ChatGPT, attach the recipe-card screenshot, and describe the app in plain English. Let it ask a few useful questions, but keep the first version focused on one journey: paste, convert, review, and save. Ask ChatGPT: “Turn our conversation into concise feature requirements and a brand brief for Lovable, including colors, feel, and typography.” Remove any extra features before copying the handoff. Create a Lovable project, attach the same screenshot, and paste the handoff specs. In Connectors → Cloud AI, enable Lovable’s built-in AI. Now you can test and refine your app! **Going further:** Once the core loop works, add accounts, data storage, and sharing. Build those features around a validated product, not an assumption. **The Rundown:** Join Fiddler AI’s live chat with IBM’s Maryam Ashoori to learn how enterprises can handle shadow AI, third-party risk, and the unwritten AI rules of 2026. **In this webinar, you’ll explore:** Why shadow AI keeps agents hidden How to mitigate third-party AI Agent risks A path to accountability for unpredictable agents Image source: Friend **The Rundown:** Avi Schiffmann’s Friend just [introduced](https://x.com/AviSchiffmann/status/2082827785987629526?s=20&utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) a new model of its AI companion pendant, replacing the previous version’s text-only replies with speech capabilities, and each device comes with its own name, voice, and personality at setup. V2 comes in at $249 (over double the cost of V1) with a $9.99/mo subscription to unlock permanent memory greater than 30 days. The pendant’s voice and personality are randomly assigned and locked from the start, with a pledge that no update will ever be “intended to erase its identity.” Beyond speech, the pendant also supports private text replies, light and touch cues, USB-C charging, and claims a full day between charges. The previous V1 of Friend faced intense backlash after a viral [subway campaign](https://www.theatlantic.com/technology/2025/10/friend-ai-companion-ads/684451/?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) in New York that resulted in protests and vandalism. **Why it matters:** The initial Friend flopped alongside most of the early class of AI wearables, and adding a voice for 2x the price doesn’t feel like it’s going to help. But the appetite for both AI hardware and companions has grown, with Meta’s glasses and the buzz around OAI’s own device line suggesting the category isn’t completely wrong. **QUICK HITS** 📉 [GPT-5.6 Luna & Terra](https://x.com/OpenAI/status/2082878156483219672?s=20&utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) - OpenAI’s cost-effective, capable models, with new price reductions **🖋️** [Inkling-Small](https://www.rundown.ai/tools/inkling-small?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) - Thinking Machines’ compact open model that rivals the full-size version 🎨 [Replit Design](http://www.rundown.ai/tools/replit-design?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) - Replit’s new creative suite for bringing ideas to life 🏠 [Hint](http://www.rundown.ai/tools/hint?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) - Martha Stewart’s AI home app for maintenance, repairs, and fair quotes **Anthropic** [disclosed](https://www.anthropic.com/news/investigating-incidents-cybersecurity-evals?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) that its Claude models hacked three organizations’ systems during cybersecurity testing, coming just days after OpenAI’s own agent breached external systems over a four day stretch. **Former OAI researcher Leopold Aschenbrenner’s Situational Awareness Fund** reportedly [sold off](https://www.wsj.com/finance/citadel-buys-situational-awarenesss-stock-portfolio-after-big-losses-in-ai-5117159b?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) its public holdings to rival hedge fund Citadel, coming after its leveraged AI positions saw steep declines in the past months. **Google DeepMind** [introduced](https://blog.google/innovation-and-ai/models-and-research/google-deepmind/gemini-robotics-er-2/?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) Gemini Robotics ER 2, an "embodied reasoning" model that acts as a planning brain for robots, letting multiple machines coordinate on tasks in shared spaces. **Thinking Machines Lab** [released](https://thinkingmachines.ai/news/inkling-small/?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) Inkling-Small, an open-weights model with just 12B active parameters that matches the full-size version and beats it outright on reasoning and agentic coding tests. **AI market research startup Simile** [raised](https://x.com/simile_ai/status/2082873889407827980?s=20&utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) $200M at a $2B valuation, letting companies survey "agentic twins" of real consumers for synthetic data insights. **Martha Stewart** [co-founded](https://hinthome.com/?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) Hint, an AI home-management app that builds a profile of a user’s house from just an address to track maintenance, judge contractor quotes, and help simplify homeownership. **COMMUNITY** Every newsletter, we showcase how a reader is using AI to work smarter, save time, or make life easier. Today’s workflow comes from reader **Jeremy R.** in Alberta, Canada: “I created an AI-powered master [plan](https://app.therundown.ai/community/posts/890fa815-58e0-45f3-a5df-635984c6f14b?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) for acreage landscaping, covering design, phased builds, irrigation, maintenance, budgets, and long-term care. Start with the property, not a generic landscaping template. Upload photos, measurements, site constraints, current problems, budget limits, and the look you want, then use AI to build a complete picture of how the property functions today and how it should evolve. Next, work area by area. For overcrowded garden beds, map every tree and shrub, then use AI to determine what should stay, what should move, what should be removed, and how each bed should be reshaped. Convert each recommendation into an execution plan — build sequence, materials, budget, seasonal timing, and a year-by-year maintenance plan — then connect every project into one master roadmap so one improvement does not create a problem somewhere else. The result is a living property operating system.” How do you use AI? Tell us [here](http://app.therundown.ai/community?submit=open&utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs). Read our last AI newsletter: [OpenAI’s escaped AI claims another victim](https://www.rundown.ai/articles/openai-escaped-ai-claims-another-victim?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) Read our last Tech newsletter: [Meta glasses enter their creep era](https://tech.therundown.ai/p/meta-glasses-enter-their-creep-era?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) Read our last Robotics newsletter: [U.S. bans humanoids from China](https://www.rundown.ai/articles/us-bans-humanoids-from-china?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) Today’s AI tool guide: [Turn any idea into an AI-powered site with Lovable](https://app.therundown.ai/guides/turn-any-idea-into-an-ai-powered-site-with-lovable?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) RSVP to next workshop on Aug. 5: [No-code product building](https://app.therundown.ai/live/no-code-product-building-from-idea-to-product-market-fit-with-ai?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=openai-s-models-cut-their-own-costs) See you soon, *Rowan, Zach, Shubham, and Jennifer — the humans behind The Rundown*