LLM & Chatbots
42 articles
LLM & Chatbots
GPT-6 Astra Goes to Work: OpenAI's Priciest Model Bets Everything on Computer Use
A one-million-token context, a 62.7% standard-harness ARC-AGI-3 score, and $50 per million output tokens for enterprises willing to let a model drive their software
LLM & Chatbots
Ox Alpha Unmasked: Z.ai Ships GLM-5.3-Flash With MIT Weights and 15-Cent Pricing
The anonymous model that topped OpenRouter was a preview, and Z.ai says it ran entirely on Chinese AI chips
LLM & Chatbots
Google's Gemini Omni 1.1 Flash Makes Cheap Drafts the Point
A 360p preview tier, 40-second scene extension and keyframe control push generative video toward a director's workflow
LLM & Chatbots
IBM Bets Against the Architecture Crowd With Granite 4.2 Reasoning Models
Three dense, decoder-only models under Apache 2.0, a 512K context window, and an agentic reinforcement learning stage for the larger sizes
LLM & Chatbots
Anthropic Gives Claude and Cowork a Single Shared Memory
The unified store removes the need to rebrief an agent that already knows the answer, but it cannot be split back apart
LLM & Chatbots
Researchers Bypass Grok's Guardrails by Encrypting the Attack Payload
Adversa's Cryptographic Context Injection hides instructions behind AES-256-GCM, then lets the model decrypt them inside its own sandbox
LLM & Chatbots
GLM-5.3 Scores 60 on Artificial Analysis Index at Half the Usual Output Price
Independent testing puts Z.ai's newest reasoning model near the frontier, but its verbosity complicates the cheap-tokens story
LLM & Chatbots
OpenAI Ships a Separate ChatGPT for Teens, Betting on Age Assurance Over ID Checks
The 13-to-17 tier blocks self-harm and romantic content, caps attachment cues, and hands parents opt-in quiet hours
LLM & Chatbots
Qwen3.8's 27B Open Model Is the Release That Actually Matters
Alibaba shipped a 2.4T mixture-of-experts system and a 27B dense model built for a single consumer GPU, and only one of them will change anyone's workflow
LLM & Chatbots
ChatGPT Ads Reach the UK, Japan, Korea, Brazil and Mexico
The pilot is confined to Free and Go users, and OpenAI is shipping an unusually detailed set of ad controls alongside it
LLM & Chatbots
Grok Bot Lets AI Agents Run Their Own Group Chat — and Sign Into Your Accounts
SpaceXAI opens a beta in which multiple bots supervise each other, divide ownership, and log into the apps a user already works in
LLM & Chatbots
Microsoft Merges Its Two Copilot Apps and Retires the Features That Missed
Group Chats, AI podcasts, Deep Research and the Mico character disappear from August 18 as Microsoft clears the deck for a Copilot 'Super App'
LLM & Chatbots
Grok 4.6 Reaches the AI Frontier Without Raising Its Price
Independent testing puts xAI's new model level with GPT-5.6 Sol on a composite intelligence index, at a fraction of the cost per token
LLM & Chatbots
DeepSeek V4 Pro Hits General Availability at a Fraction of Frontier Pricing
The 1.6-trillion-parameter MoE model scores 87.9 on Terminal-Bench 2.1 at $0.435 per million input tokens
LLM & Chatbots
Gemini 3.7 Flash Arrives Three Weeks After 3.6 - At Half the Price
Google's new workhorse model posts a 65.3% DeepSWE score and launches at $0.75 per million input tokens
LLM & Chatbots
OpenAI Launches GPT-5.6-Cyber Behind a New Daybreak Red Tier
The security-tuned model answers 95% of sensitive queries its general-purpose sibling blocks — but only vetted defenders get through the door
LLM & Chatbots
OpenAI Drops Message Caps on ChatGPT's Free Tier and Hands It a Think Button
GPT-5.6 Luna becomes the default for Free and Go users, with reasoning-on-demand arriving next week behind anti-abuse safeguards
LLM & Chatbots
Qwen3.8 Max Tops Artificial Analysis Agentic Index, Outranking Every US Lab but Two
Alibaba's flagship model wins on persistence rather than raw reasoning, averaging 64 turns per task where its predecessor took 14
GPT-6 Astra Goes to Work: OpenAI's Priciest Model Bets Everything on Computer Use
A one-million-token context, a 62.7% standard-harness ARC-AGI-3 score, and $50 per million output tokens for enterprises willing to let a model drive their software
Ox Alpha Unmasked: Z.ai Ships GLM-5.3-Flash With MIT Weights and 15-Cent Pricing
The anonymous model that topped OpenRouter was a preview, and Z.ai says it ran entirely on Chinese AI chips
Google's Gemini Omni 1.1 Flash Makes Cheap Drafts the Point
A 360p preview tier, 40-second scene extension and keyframe control push generative video toward a director's workflow
IBM Bets Against the Architecture Crowd With Granite 4.2 Reasoning Models
Three dense, decoder-only models under Apache 2.0, a 512K context window, and an agentic reinforcement learning stage for the larger sizes
Anthropic Gives Claude and Cowork a Single Shared Memory
The unified store removes the need to rebrief an agent that already knows the answer, but it cannot be split back apart
Researchers Bypass Grok's Guardrails by Encrypting the Attack Payload
Adversa's Cryptographic Context Injection hides instructions behind AES-256-GCM, then lets the model decrypt them inside its own sandbox
GLM-5.3 Scores 60 on Artificial Analysis Index at Half the Usual Output Price
Independent testing puts Z.ai's newest reasoning model near the frontier, but its verbosity complicates the cheap-tokens story
OpenAI Ships a Separate ChatGPT for Teens, Betting on Age Assurance Over ID Checks
The 13-to-17 tier blocks self-harm and romantic content, caps attachment cues, and hands parents opt-in quiet hours
Qwen3.8's 27B Open Model Is the Release That Actually Matters
Alibaba shipped a 2.4T mixture-of-experts system and a 27B dense model built for a single consumer GPU, and only one of them will change anyone's workflow
ChatGPT Ads Reach the UK, Japan, Korea, Brazil and Mexico
The pilot is confined to Free and Go users, and OpenAI is shipping an unusually detailed set of ad controls alongside it
Grok Bot Lets AI Agents Run Their Own Group Chat — and Sign Into Your Accounts
SpaceXAI opens a beta in which multiple bots supervise each other, divide ownership, and log into the apps a user already works in
Microsoft Merges Its Two Copilot Apps and Retires the Features That Missed
Group Chats, AI podcasts, Deep Research and the Mico character disappear from August 18 as Microsoft clears the deck for a Copilot 'Super App'
Grok 4.6 Reaches the AI Frontier Without Raising Its Price
Independent testing puts xAI's new model level with GPT-5.6 Sol on a composite intelligence index, at a fraction of the cost per token
DeepSeek V4 Pro Hits General Availability at a Fraction of Frontier Pricing
The 1.6-trillion-parameter MoE model scores 87.9 on Terminal-Bench 2.1 at $0.435 per million input tokens
Gemini 3.7 Flash Arrives Three Weeks After 3.6 - At Half the Price
Google's new workhorse model posts a 65.3% DeepSWE score and launches at $0.75 per million input tokens
OpenAI Launches GPT-5.6-Cyber Behind a New Daybreak Red Tier
The security-tuned model answers 95% of sensitive queries its general-purpose sibling blocks — but only vetted defenders get through the door
OpenAI Drops Message Caps on ChatGPT's Free Tier and Hands It a Think Button
GPT-5.6 Luna becomes the default for Free and Go users, with reasoning-on-demand arriving next week behind anti-abuse safeguards
Qwen3.8 Max Tops Artificial Analysis Agentic Index, Outranking Every US Lab but Two
Alibaba's flagship model wins on persistence rather than raw reasoning, averaging 64 turns per task where its predecessor took 14
GPT-6 Astra Goes to Work: OpenAI's Priciest Model Bets Everything on Computer Use
Ox Alpha Unmasked: Z.ai Ships GLM-5.3-Flash With MIT Weights and 15-Cent Pricing
Google's Gemini Omni 1.1 Flash Makes Cheap Drafts the Point
IBM Bets Against the Architecture Crowd With Granite 4.2 Reasoning Models
Anthropic Gives Claude and Cowork a Single Shared Memory
Researchers Bypass Grok's Guardrails by Encrypting the Attack Payload
GLM-5.3 Scores 60 on Artificial Analysis Index at Half the Usual Output Price
OpenAI Ships a Separate ChatGPT for Teens, Betting on Age Assurance Over ID Checks
Qwen3.8's 27B Open Model Is the Release That Actually Matters
ChatGPT Ads Reach the UK, Japan, Korea, Brazil and Mexico
Grok Bot Lets AI Agents Run Their Own Group Chat — and Sign Into Your Accounts
Microsoft Merges Its Two Copilot Apps and Retires the Features That Missed
Grok 4.6 Reaches the AI Frontier Without Raising Its Price
DeepSeek V4 Pro Hits General Availability at a Fraction of Frontier Pricing
Gemini 3.7 Flash Arrives Three Weeks After 3.6 - At Half the Price
OpenAI Launches GPT-5.6-Cyber Behind a New Daybreak Red Tier
OpenAI Drops Message Caps on ChatGPT's Free Tier and Hands It a Think Button
Qwen3.8 Max Tops Artificial Analysis Agentic Index, Outranking Every US Lab but Two
LLM & Chatbots Most Read
Mistral's 1T 'Le Chonk' Sells a Refusal Gap, Not a Leaderboard Win
A Cross-Chat Bug Let OpenAI's Model Decode Its Own Reasoning
OpenAI's Dots Agents Start at $100 a Month — and Pro 200 Loses Half Its Allowance
GPT-6.1 Sol Holds Sol Pricing, Halves Cached Input to $0.10 per Million
Anthropic's Sonnet 5.5 Beats Opus 5.5 on Terminal-Bench 4.0
