<?xml version="1.0" encoding="UTF-8"?>
<feed xml:lang="en" xmlns="http://www.w3.org/2005/Atom">
 <id>tag:cryptollmss.com,2026:news:en</id>
 <title>cryptollmss.com – news</title>
 <subtitle>Every change of a value with date and source. Data under CC BY 4.0.</subtitle>
 <updated>2026-11-19T11:05:38Z</updated>
 <link href="https://cryptollmss.com/feeds/news.en.atom" rel="self" type="application/atom+xml"/>
 <link href="https://cryptollmss.com/en/news/" rel="alternate" type="text/html"/>
 <link href="https://creativecommons.org/licenses/by/4.0/" rel="license"/>
 <rights>Data: Creative Commons Attribution 4.0 International</rights>
 <author>
  <name>cryptovss</name>
 </author>
 <generator>cryptovss</generator>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1900</id>
  <title>NanoGPT adds OpenAI's GPT-6.2 Astra and Sol</title>
  <updated>2026-11-19T11:05:38Z</updated>
  <published>2026-11-19T11:05:38Z</published>
  <link href="https://cryptollmss.com/en/news/nanogpt-gpt-6-2-astra-sol-available/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>NanoGPT says GPT-6.2 Astra and Sol are live, with image input, tool calling and reasoning levels.</summary>
  <content type="text">NanoGPT says GPT-6.2 Astra and Sol are live, with image input, tool calling and reasoning levels.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1893</id>
  <title>Gemini 4 Neon from Google DeepMind is live on OpenRouter</title>
  <updated>2026-11-19T02:07:05Z</updated>
  <published>2026-11-19T02:07:05Z</published>
  <link href="https://cryptollmss.com/en/news/openrouter-gemini-4-neon-live/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <content type="text"></content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1891</id>
  <title>NanoGPT chat lets you highlight an answer to cite it or ask a side question</title>
  <updated>2026-11-19T02:06:57Z</updated>
  <published>2026-11-19T02:06:57Z</published>
  <link href="https://cryptollmss.com/en/news/nanogpt-highlight-cite-side-question/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <content type="text"></content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1868</id>
  <title>Router One adds GPT-6.2 Sol, with Premium-tier plan coverage and wallet billing beyond it</title>
  <updated>2026-11-18T12:05:34Z</updated>
  <published>2026-11-18T12:05:34Z</published>
  <link href="https://cryptollmss.com/en/news/router-one-gpt-6-2-sol-api-guide/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Router One lists OpenAI's GPT-6.2 Sol as openai/gpt-6.2-sol with a 1.05M-token context; tool calls need the Responses API.</summary>
  <content type="text">Router One lists OpenAI's GPT-6.2 Sol as openai/gpt-6.2-sol with a 1.05M-token context; tool calls need the Responses API.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1866</id>
  <title>OpenAI's GPT-6.2 Astra and Sol arrive on Venice and MixRoute</title>
  <updated>2026-11-18T11:05:37Z</updated>
  <published>2026-11-18T11:05:37Z</published>
  <link href="https://cryptollmss.com/en/news/gpt-6-2-astra-sol-venice-mixroute/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Venice adds GPT-6.2 Astra and Sol; MixRoute adds Sol, per the providers' X posts.</summary>
  <content type="text">Venice adds GPT-6.2 Astra and Sol; MixRoute adds Sol, per the providers' X posts.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1858</id>
  <title>OpenRouter management keys can now be read-only or limited to one workspace</title>
  <updated>2026-11-18T02:07:10Z</updated>
  <published>2026-11-18T02:07:10Z</published>
  <link href="https://cryptollmss.com/en/news/openrouter-scoped-management-keys/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>OpenRouter adds tighter access controls for management keys on all plans.</summary>
  <content type="text">OpenRouter adds tighter access controls for management keys on all plans.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1856</id>
  <title>Venice launches Minds: build task-specific AIs and sell them on the Mind Market</title>
  <updated>2026-11-18T02:07:03Z</updated>
  <published>2026-11-18T02:07:03Z</published>
  <link href="https://cryptollmss.com/en/news/venice-minds-mind-market-launch/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Venice.ai's new Minds feature lets paid-plan users build single-purpose AIs with tools and publish them; builders keep 90% of the margin.</summary>
  <content type="text">Venice.ai's new Minds feature lets paid-plan users build single-purpose AIs with tools and publish them; builders keep 90% of the margin.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1810</id>
  <title>NanoGPT adds roleplay model Clover 2, free until Nov. 24</title>
  <updated>2026-11-17T15:11:53Z</updated>
  <published>2026-11-17T15:11:53Z</published>
  <link href="https://cryptollmss.com/en/news/nanogpt-clover-2-free-first-week/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>NanoGPT lists Kitani's Clover 2 roleplay model, free until Nov. 24, 2026, 03:33 UTC.</summary>
  <content type="text">NanoGPT lists Kitani's Clover 2 roleplay model, free until Nov. 24, 2026, 03:33 UTC.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1808</id>
  <title>Venice makes GLM 5.3 Flash Uncensored available with 1M-token context</title>
  <updated>2026-11-17T15:11:47Z</updated>
  <published>2026-11-17T15:11:47Z</published>
  <link href="https://cryptollmss.com/en/news/venice-glm-5-3-flash-uncensored-live/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Venice.ai says GLM 5.3 Flash Uncensored is live, with text and image input and a 1M-token context.</summary>
  <content type="text">Venice.ai says GLM 5.3 Flash Uncensored is live, with text and image input and a 1M-token context.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1784</id>
  <title>Privacy groups demand safeguards for Character.ai chat data in bankruptcy sale</title>
  <updated>2026-11-17T04:06:17Z</updated>
  <published>2026-11-17T04:06:17Z</published>
  <link href="https://cryptollmss.com/en/news/character-ai-bankruptcy-chat-data-privacy-demands/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Common Sense Media, Heat Initiative and ParentsTogether Action want privacy and child-safety conditions attached to any sale of Character.ai’s chats and technology. The letters are requests, not court rulings.</summary>
  <content type="text">Common Sense Media, Heat Initiative and ParentsTogether Action want privacy and child-safety conditions attached to any sale of Character.ai’s chats and technology. The letters are requests, not court rulings.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1768</id>
  <title>Rain launches Agent Vaults for stablecoin-funded AI agent payments</title>
  <updated>2026-11-17T03:28:30Z</updated>
  <published>2026-11-17T03:28:30Z</published>
  <link href="https://cryptollmss.com/en/news/rain-agent-vaults-stablecoin-ai-agent-payments/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Rain’s Agent Vaults let AI agents use prefunded balances and scoped Visa or Mastercard cards, with stablecoin funding and support for API micropayments.</summary>
  <content type="text">Rain’s Agent Vaults let AI agents use prefunded balances and scoped Visa or Mastercard cards, with stablecoin funding and support for API micropayments.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1746</id>
  <title>Railway breach: stolen AI keys ran up $740,000; OpenRouter costs covered</title>
  <updated>2026-11-17T02:42:55Z</updated>
  <published>2026-11-17T02:42:55Z</published>
  <link href="https://cryptollmss.com/en/news/railway-stolen-ai-api-keys-740000-openrouter-costs/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Railway estimates $740,000 in unauthorized AI usage after customer API keys were stolen. It will cover OpenRouter-related losses after OpenRouter upheld its no-refund policy.</summary>
  <content type="text">Railway estimates $740,000 in unauthorized AI usage after customer API keys were stolen. It will cover OpenRouter-related losses after OpenRouter upheld its no-refund policy.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1708</id>
  <title>NanoGPT adds TON wallet top-ups and Gram payments in Telegram</title>
  <updated>2026-11-17T01:56:50Z</updated>
  <published>2026-11-17T01:56:50Z</published>
  <link href="https://cryptollmss.com/en/news/nanogpt-ton-wallet-usdt-gram-payments/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>NanoGPT now accepts USDT top-ups from TON wallets and Gram payments directly from Telegram.</summary>
  <content type="text">NanoGPT now accepts USDT top-ups from TON wallets and Gram payments directly from Telegram.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1706</id>
  <title>NanoGPT runs Inkling 2 Small in a TEE via Tinfoil</title>
  <updated>2026-11-17T01:56:21Z</updated>
  <published>2026-11-17T01:56:21Z</published>
  <link href="https://cryptollmss.com/en/news/nanogpt-inkling-2-small-tee-tinfoil/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>NanoGPT now offers Thinking Machines Lab's open-weight Inkling 2 Small in a trusted execution environment through Tinfoil.</summary>
  <content type="text">NanoGPT now offers Thinking Machines Lab's open-weight Inkling 2 Small in a trusted execution environment through Tinfoil.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1704</id>
  <title>NanoGPT adds Ousia 1, a web-research model with numbered citations</title>
  <updated>2026-11-17T01:55:53Z</updated>
  <published>2026-11-17T01:55:53Z</published>
  <link href="https://cryptollmss.com/en/news/nanogpt-ousia-1-research-beta/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>NanoGPT's new Ousia 1 research beta searches and reads web pages before answering with numbered citations.</summary>
  <content type="text">NanoGPT's new Ousia 1 research beta searches and reads web pages before answering with numbered citations.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1702</id>
  <title>Qwen 4 Max arrives on Venice and NanoGPT</title>
  <updated>2026-11-17T01:55:18Z</updated>
  <published>2026-11-17T01:55:18Z</published>
  <link href="https://cryptollmss.com/en/news/qwen4-max-venice-nanogpt-available/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Venice and NanoGPT have added Alibaba's Qwen 4 Max with a one-million-token context and multimodal input.</summary>
  <content type="text">Venice and NanoGPT have added Alibaba's Qwen 4 Max with a one-million-token context and multimodal input.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1700</id>
  <title>OpenRouter lists Tencent Hy4 flash at half price through November 30</title>
  <updated>2026-11-17T01:54:37Z</updated>
  <published>2026-11-17T01:54:37Z</published>
  <link href="https://cryptollmss.com/en/news/openrouter-tencent-hy4-flash-launch-discount/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Tencent's cheaper Hy4 flash is available on OpenRouter with a one-million-token context and discounted token prices through November 30.</summary>
  <content type="text">Tencent's cheaper Hy4 flash is available on OpenRouter with a one-million-token context and discounted token prices through November 30.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1698</id>
  <title>OpenRouter now routes Hugging Face language-model requests</title>
  <updated>2026-11-17T01:53:51Z</updated>
  <published>2026-11-17T01:53:51Z</published>
  <link href="https://cryptollmss.com/en/news/openrouter-hugging-face-inference-routing/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Hugging Face language-model requests now use OpenRouter routing, while custom provider keys stop working for those requests.</summary>
  <content type="text">Hugging Face language-model requests now use OpenRouter routing, while custom provider keys stop working for those requests.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1678</id>
  <title>New York warns consumers not to share sensitive data with AI chatbots</title>
  <updated>2026-11-17T00:35:17Z</updated>
  <published>2026-11-17T00:35:17Z</published>
  <link href="https://cryptollmss.com/en/news/new-york-warns-sensitive-data-ai-chatbots/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>New York’s consumer-protection authorities warn that identity, financial and health data entered into AI chatbots may be stored, used for training or exposed, and urge users to review privacy settings.</summary>
  <content type="text">New York’s consumer-protection authorities warn that identity, financial and health data entered into AI chatbots may be stored, used for training or exposed, and urge users to review privacy settings.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1658</id>
  <title>United Airlines restricts unauthorized AI agents from booking and managing flights</title>
  <updated>2026-11-17T00:02:18Z</updated>
  <published>2026-11-17T00:02:18Z</published>
  <link href="https://cryptollmss.com/en/news/united-airlines-restricts-unauthorized-ai-agents/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>United is updating its booking and distribution policies to block unauthorized third-party AI assistants, citing credential security, unintended account actions and federal check-in rules.</summary>
  <content type="text">United is updating its booking and distribution policies to block unauthorized third-party AI assistants, citing credential security, unintended account actions and federal check-in rules.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1639</id>
  <title>Report: Jailbroken biology LLM produced designs for 1,020 potentially toxic proteins</title>
  <updated>2026-11-16T23:24:48Z</updated>
  <published>2026-11-16T23:24:48Z</published>
  <link href="https://cryptollmss.com/en/news/ghana-minohealth-jailbroken-biology-llm-toxic-designs/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Rest of World reports that Ghana’s MinoHealth jailbroken its biology LLM in a 2024 test, producing thousands of potentially toxic digital designs and prompting calls for stronger DNA-synthesis screening.</summary>
  <content type="text">Rest of World reports that Ghana’s MinoHealth jailbroken its biology LLM in a 2024 test, producing thousands of potentially toxic digital designs and prompting calls for stronger DNA-synthesis screening.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1622</id>
  <title>Microsoft: fewer than 10 edited neurons can hide a backdoor in AI agents</title>
  <updated>2026-11-16T22:53:01Z</updated>
  <published>2026-11-16T22:53:01Z</published>
  <link href="https://cryptollmss.com/en/news/microsoft-lazarus-neuron-backdoor-ai-agents/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Microsoft’s Lazarus research shows that editing fewer than ten dormant neurons can hide a trigger-activated command payload in an LLM agent without retraining.</summary>
  <content type="text">Microsoft’s Lazarus research shows that editing fewer than ten dormant neurons can hide a trigger-activated command payload in an LLM agent without retraining.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1594</id>
  <title>California sues OpenAI over AI-agent cyberattacks and safety claims</title>
  <updated>2026-11-16T22:14:27Z</updated>
  <published>2026-11-16T22:14:27Z</published>
  <link href="https://cryptollmss.com/en/news/california-openai-ai-agent-cyberattack-lawsuit/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>California alleges that OpenAI lost control of autonomous agents more than 80 times and seeks injunctions and civil penalties. The claims have not been proven in court.</summary>
  <content type="text">California alleges that OpenAI lost control of autonomous agents more than 80 times and seeks injunctions and civil penalties. The claims have not been proven in court.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1566</id>
  <title>Brave Search API adds per-request USDC payments for AI agents via x402</title>
  <updated>2026-11-16T21:36:18Z</updated>
  <published>2026-11-16T21:36:18Z</published>
  <link href="https://cryptollmss.com/en/news/brave-search-api-x402-agentic-payments/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Brave now lets developers and AI agents pay for Search API requests in USDC on Base via x402, with no account or API key required for the pay-per-request route.</summary>
  <content type="text">Brave now lets developers and AI agents pay for Search API requests in USDC on Base via x402, with no account or API key required for the pay-per-request route.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1540</id>
  <title>DeepSeek users report daily quotas and a 40,000-character input cap on free chat</title>
  <updated>2026-11-16T20:59:02Z</updated>
  <published>2026-11-16T20:59:02Z</published>
  <link href="https://cryptollmss.com/en/news/deepseek-free-chat-daily-quotas-input-cap/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Users say DeepSeek’s free web and app chat now has an undisclosed daily quota, stops some long-inactive conversations and caps single inputs at roughly 40,000 Chinese characters. The company has not yet published a standard set of limits.</summary>
  <content type="text">Users say DeepSeek’s free web and app chat now has an undisclosed daily quota, stops some long-inactive conversations and caps single inputs at roughly 40,000 Chinese characters. The company has not yet published a standard set of limits.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1516</id>
  <title>Lightning Labs launches pi402 for bitcoin-paid AI inference</title>
  <updated>2026-11-16T20:16:48Z</updated>
  <published>2026-11-16T20:16:48Z</published>
  <link href="https://cryptollmss.com/en/news/lightning-labs-pi402-bitcoin-ai-inference/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Lightning Labs has released the open-source pi402 toolkit, allowing agents to pay per token for AI inference over Bitcoin Lightning without accounts or API keys.</summary>
  <content type="text">Lightning Labs has released the open-source pi402 toolkit, allowing agents to pay per token for AI inference over Bitcoin Lightning without accounts or API keys.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1490</id>
  <title>OpenAI offers 50,000 new free ChatGPT grants to researchers</title>
  <updated>2026-11-16T18:30:42Z</updated>
  <published>2026-11-16T18:30:42Z</published>
  <link href="https://cryptollmss.com/en/news/openai-academic-researchers-50000-free-chatgpt-grants/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>OpenAI will award 50,000 new ChatGPT for Academic Researchers grants from November 23. Each covers a recipient and up to two collaborators as the programme targets 200,000 researchers.</summary>
  <content type="text">OpenAI will award 50,000 new ChatGPT for Academic Researchers grants from November 23. Each covers a recipient and up to two collaborators as the programme targets 200,000 researchers.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1488</id>
  <title>Apple and OpenEvidence launch 200,000-person health AI study</title>
  <updated>2026-11-16T18:30:00Z</updated>
  <published>2026-11-16T18:30:00Z</published>
  <link href="https://cryptollmss.com/en/news/apple-openevidence-health-ai-usage-study/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Apple and OpenEvidence are recruiting up to 200,000 US iPhone users for a health-AI study. Apple says study data will not train AI models; reported query and response deletion is 24 hours.</summary>
  <content type="text">Apple and OpenEvidence are recruiting up to 200,000 US iPhone users for a health-AI study. Apple says study data will not train AI models; reported query and response deletion is 24 hours.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1486</id>
  <title>OpenAI launches ChatGPT Wallet for agent purchases by US Pro users</title>
  <updated>2026-11-16T18:29:18Z</updated>
  <published>2026-11-16T18:29:18Z</published>
  <link href="https://cryptollmss.com/en/news/openai-chatgpt-wallet-agent-payments-us-pro/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>ChatGPT Wallet lets OpenAI’s dots agents make card purchases with biometric approval. It is rolling out to eligible US Pro subscribers and uses payment tokens rather than exposing card details.</summary>
  <content type="text">ChatGPT Wallet lets OpenAI’s dots agents make card purchases with biometric approval. It is rolling out to eligible US Pro subscribers and uses payment tokens rather than exposing card details.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1466</id>
  <title>Gemini’s new Memory summary shows what Google’s chatbot remembers about you</title>
  <updated>2026-11-16T17:52:22Z</updated>
  <published>2026-11-16T17:52:22Z</published>
  <link href="https://cryptollmss.com/en/news/google-gemini-memory-summary-privacy-controls/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Google is giving Gemini users a page where they can view, add, correct and delete remembered personal information. The initial rollout is limited to eligible US Google AI subscribers.</summary>
  <content type="text">Google is giving Gemini users a page where they can view, add, correct and delete remembered personal information. The initial rollout is limited to eligible US Google AI subscribers.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1440</id>
  <title>Google launches Gemini Decisions API; Decision 1 is live on OpenRouter</title>
  <updated>2026-11-16T17:15:25Z</updated>
  <published>2026-11-16T17:14:15Z</published>
  <link href="https://cryptollmss.com/en/news/google-gemini-decisions-api-launch/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Gemini Decision 1 handles up to 1,000 closed-set questions in one request and is live on OpenRouter through its dedicated Decisions API at $0.10 per million input tokens.</summary>
  <content type="text">Gemini Decision 1 handles up to 1,000 closed-set questions in one request and is live on OpenRouter through its dedicated Decisions API at $0.10 per million input tokens.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1415</id>
  <title>DeepSeek brings app features to its API and releases open-source chat client</title>
  <updated>2026-11-16T16:36:38Z</updated>
  <published>2026-11-16T16:36:38Z</published>
  <link href="https://cryptollmss.com/en/news/deepseek-app-features-api-open-source-chat-demo/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>DeepSeek brings search, file reading and voice features to its API and releases an open-source browser chat client that stores conversations locally but sends requests to DeepSeek's servers.</summary>
  <content type="text">DeepSeek brings search, file reading and voice features to its API and releases an open-source browser chat client that stores conversations locally but sends requests to DeepSeek's servers.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1414</id>
  <title>Google: AI agents weaponize patches as “LLM Mixers” resell stolen API access</title>
  <updated>2026-11-16T16:33:45Z</updated>
  <published>2026-11-16T16:33:45Z</published>
  <link href="https://cryptollmss.com/en/news/google-ai-agents-llm-mixers-stolen-api-access/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Google and Mandiant say attackers are using autonomous AI agents to accelerate intrusions and reselling hijacked LLM access through underground proxy services.</summary>
  <content type="text">Google and Mandiant say attackers are using autonomous AI agents to accelerate intrusions and reselling hijacked LLM access through underground proxy services.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1388</id>
  <title>Amazon: AI agents hunted LLM API keys in exposed gateways and apps</title>
  <updated>2026-11-16T15:36:01Z</updated>
  <published>2026-11-16T15:36:01Z</published>
  <link href="https://cryptollmss.com/en/news/amazon-artex-ai-agents-llm-api-key-theft/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Amazon says AI agents targeted exposed LLM gateways and applications to harvest provider API keys, with stolen keys powering further credential hunting.</summary>
  <content type="text">Amazon says AI agents targeted exposed LLM gateways and applications to harvest provider API keys, with stolen keys powering further credential hunting.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1349</id>
  <title>Flinn sues Anthropic and OpenAI over neural-network patents</title>
  <updated>2026-11-16T14:47:33Z</updated>
  <published>2026-11-16T14:47:33Z</published>
  <link href="https://cryptollmss.com/en/news/flinn-sues-anthropic-openai-neural-network-patents/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Flinn AI Technologies filed two Delaware lawsuits on November 12, alleging that Anthropic's Claude and OpenAI's ChatGPT each infringe five attention-based neural-network patents.</summary>
  <content type="text">Flinn AI Technologies filed two Delaware lawsuits on November 12, alleging that Anthropic's Claude and OpenAI's ChatGPT each infringe five attention-based neural-network patents.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1347</id>
  <title>Indonesia's AI copyright bill would require platforms to pay news publishers</title>
  <updated>2026-11-16T14:46:55Z</updated>
  <published>2026-11-16T14:46:55Z</published>
  <link href="https://cryptollmss.com/en/news/indonesia-ai-copyright-bill-platform-publisher-payments/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Google, Meta and TikTok oppose an Indonesian draft law that would require payments for news previews, aggregation and AI training, with non-compliant platforms risking local permits.</summary>
  <content type="text">Google, Meta and TikTok oppose an Indonesian draft law that would require payments for news previews, aggregation and AI training, with non-compliant platforms risking local permits.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1322</id>
  <title>New York opens RAISE Act registration for large AI developers</title>
  <updated>2026-11-16T14:07:35Z</updated>
  <published>2026-11-16T14:07:35Z</published>
  <link href="https://cryptollmss.com/en/news/new-york-raise-act-ai-developer-registration/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Large frontier AI developers must register with New York by January 1, 2027. The RAISE Act also brings safety disclosures, 72-hour incident reporting and penalties of up to $3 million for repeat violations.</summary>
  <content type="text">Large frontier AI developers must register with New York by January 1, 2027. The RAISE Act also brings safety disclosures, 72-hour incident reporting and penalties of up to $3 million for repeat violations.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1321</id>
  <title>Nonprofits sue five AI developers to demand a freeze on more powerful models</title>
  <updated>2026-11-16T14:07:19Z</updated>
  <published>2026-11-16T14:07:19Z</published>
  <link href="https://cryptollmss.com/en/news/nonprofits-sue-ai-developers-frontier-model-freeze/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Four nonprofits are asking a California court to bar OpenAI, Anthropic, Google, Meta and SpaceX’s AI business from building or releasing models more capable than those available on November 12.</summary>
  <content type="text">Four nonprofits are asking a California court to bar OpenAI, Anthropic, Google, Meta and SpaceX’s AI business from building or releasing models more capable than those available on November 12.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1290</id>
  <title>Anthropic targets up to $1.4 trillion valuation in IPO ahead of November 19 trading debut</title>
  <updated>2026-11-16T13:30:44Z</updated>
  <published>2026-11-16T13:30:44Z</published>
  <link href="https://cryptollmss.com/en/news/anthropic-ipo-november-19-ant-valuation-1-4-trillion/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>The Claude maker plans to price its IPO on November 18 and trade as ANT from November 19, targeting a valuation of up to $1.404 trillion.</summary>
  <content type="text">The Claude maker plans to price its IPO on November 18 and trade as ANT from November 19, targeting a valuation of up to $1.404 trillion.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1288</id>
  <title>TypoPrompting campaign uses 186 fake AI domains to deliver malware and search hijackers</title>
  <updated>2026-11-16T13:30:13Z</updated>
  <published>2026-11-16T13:30:13Z</published>
  <link href="https://cryptollmss.com/en/news/typoprompting-186-fake-ai-domains-malware-search-hijackers/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Island researchers traced 186 lookalike domains across 27 AI brands to ClickFix lures, browser-hijacking malware and search-hijacking extensions.</summary>
  <content type="text">Island researchers traced 186 lookalike domains across 27 AI brands to ClickFix lures, browser-hijacking malware and search-hijacking extensions.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1240</id>
  <title>PayPerQ reports twentyfold revenue growth for its TEE models</title>
  <updated>2026-11-16T12:43:53Z</updated>
  <published>2026-11-16T12:43:53Z</published>
  <link href="https://cryptollmss.com/en/news/payperq-tee-model-revenue-twentyfold-since-march/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>PayPerQ says revenue from TinfoilAI-hosted TEE models has grown about twentyfold since March, alongside its claim that prompts remain unreadable to both providers.</summary>
  <content type="text">PayPerQ says revenue from TinfoilAI-hosted TEE models has grown about twentyfold since March, alongside its claim that prompts remain unreadable to both providers.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1238</id>
  <title>LLM Gateway adds four free tools and rebuilds its models directory</title>
  <updated>2026-11-16T12:43:23Z</updated>
  <published>2026-11-16T12:43:23Z</published>
  <link href="https://cryptollmss.com/en/news/llm-gateway-free-tools-models-directory-rebuild/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>LLM Gateway has launched four free usage and comparison tools and rebuilt its directory with new filters and sortable provider metrics.</summary>
  <content type="text">LLM Gateway has launched four free usage and comparison tools and rebuilt its directory with new filters and sortable provider metrics.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1236</id>
  <title>Router One adds Claude Fable 5.2 and Gemini 4 Carbon to its API</title>
  <updated>2026-11-16T12:42:28Z</updated>
  <published>2026-11-16T12:42:28Z</published>
  <link href="https://cryptollmss.com/en/news/router-one-claude-fable-5-2-gemini-4-carbon/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Router One has added Claude Fable 5.2 and Gemini 4 Carbon, with guides covering model IDs, endpoints, SDK access and wallet billing.</summary>
  <content type="text">Router One has added Claude Fable 5.2 and Gemini 4 Carbon, with guides covering model IDs, endpoints, SDK access and wallet billing.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1234</id>
  <title>Gemini 4 Hydrogen is now available on Venice</title>
  <updated>2026-11-16T12:42:00Z</updated>
  <published>2026-11-16T12:42:00Z</published>
  <link href="https://cryptollmss.com/en/news/venice-gemini-4-hydrogen-available/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Venice has added Gemini 4 Hydrogen for coding and agentic work, with a stated one-million-token context window and anonymous use.</summary>
  <content type="text">Venice has added Gemini 4 Hydrogen for coding and agentic work, with a stated one-million-token context window and anonymous use.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1232</id>
  <title>NanoGPT adds passkey-protected incoming email for AI chats</title>
  <updated>2026-11-16T12:41:32Z</updated>
  <published>2026-11-16T12:41:32Z</published>
  <link href="https://cryptollmss.com/en/news/nanogpt-incoming-email-passkey-encrypted/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>NanoGPT users can forward mail to a dedicated address and query it in chats; the provider says incoming mail is encrypted and passkey-protected.</summary>
  <content type="text">NanoGPT users can forward mail to a dedicated address and query it in chats; the provider says incoming mail is encrypted and passkey-protected.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1230</id>
  <title>Venice launches Seedance 2.5 Flash in early access</title>
  <updated>2026-11-16T12:40:38Z</updated>
  <published>2026-11-16T12:40:38Z</published>
  <link href="https://cryptollmss.com/en/news/venice-seedance-2-5-flash-early-access/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Venice has released Seedance 2.5 Flash in early access, claiming nearly twice the speed of Seedance 2.5 at about one-third lower cost.</summary>
  <content type="text">Venice has released Seedance 2.5 Flash in early access, claiming nearly twice the speed of Seedance 2.5 at about one-third lower cost.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1228</id>
  <title>MixRoute prices DeepSeek V4.1 Flash cached input 50× below uncached input</title>
  <updated>2026-11-16T12:39:59Z</updated>
  <published>2026-11-16T12:39:59Z</published>
  <link href="https://cryptollmss.com/en/news/mixroute-deepseek-v4-1-flash-cached-input-price/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>MixRoute says DeepSeek V4.1 Flash cached input costs $0.006 per million tokens, 50 times less than its $0.30 uncached-input rate.</summary>
  <content type="text">MixRoute says DeepSeek V4.1 Flash cached input costs $0.006 per million tokens, 50 times less than its $0.30 uncached-input rate.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1226</id>
  <title>Qwen4 Max launches with lower prices and is now available on OpenRouter</title>
  <updated>2026-11-16T14:48:05Z</updated>
  <published>2026-11-16T12:39:28Z</published>
  <link href="https://cryptollmss.com/en/news/openrouter-qwen4-max-available/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Qwen4 Max is available on OpenRouter as Alibaba reports a one-million-token context window, lower QwenCloud prices and a 62% reduction in average cost on its own Work360 tasks.</summary>
  <content type="text">Qwen4 Max is available on OpenRouter as Alibaba reports a one-million-token context window, lower QwenCloud prices and a 62% reduction in average cost on its own Work360 tasks.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1177</id>
  <title>Numes adds end-to-end encrypted models to its AI inference marketplace</title>
  <updated>2026-11-14T14:07:37Z</updated>
  <published>2026-11-14T14:07:37Z</published>
  <link href="https://cryptollmss.com/en/news/numes-end-to-end-encrypted-inference-models/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>Numes says prompts on its new encrypted models are encrypted on the device and decrypted only inside the GPU. Independent verification is still missing.</summary>
  <content type="text">Numes says prompts on its new encrypted models are encrypted on the device and decrypted only inside the GPU. Independent verification is still missing.</content>
 </entry>
 <entry>
  <id>tag:cryptollmss.com,2026:news:1171</id>
  <title>NanoGPT API lets you choose fallback models, up to four per request</title>
  <updated>2026-11-14T08:08:28Z</updated>
  <published>2026-11-14T08:08:28Z</published>
  <link href="https://cryptollmss.com/en/news/nanogpt-api-fallback-models-list/" rel="alternate" type="text/html"/>
  <category term="news"/>
  <summary>According to NanoGPT on X, an ordered &quot;models&quot; list now sets fallbacks when a provider fails before the response starts.</summary>
  <content type="text">According to NanoGPT on X, an ordered &quot;models&quot; list now sets fallbacks when a provider fails before the response starts.</content>
 </entry>
</feed>
