Models

Releases, benchmarks, and analysis of the latest AI models from frontier labs.

Anthropic Debuts on Nasdaq Amid Safety Research Push
Models

Anthropic Debuts on Nasdaq Amid Safety Research Push

AI safety-focused company begins trading as it maintains research commitment

Anthropic Shifts to Multi-Model Claude Strategy at IPO
Models

Anthropic Shifts to Multi-Model Claude Strategy at IPO

Anthropic IPO marks shift to multi-model Claude strategy

llama.cpp 1.0 Marks Inflection Point for Local AI
Models

llama.cpp 1.0 Marks Inflection Point for Local AI

Infrastructure milestones strengthen open-source LLM ecosystem

Agent Frameworks Push Open Interoperability Standard
Models

Agent Frameworks Push Open Interoperability Standard

ECC and Google Labs align on standards for agent capabilities

Claude embeds in Microsoft 365 for 400M enterprise users
Models

Claude embeds in Microsoft 365 for 400M enterprise users

Anthropic's largest platform deal brings Claude to M365 Copilot starting Dec. 9 in UK

Local Open Source Coding Models Challenge API Economics
Models

Local Open Source Coding Models Challenge API Economics

Free, self-hosted coding LLMs gain ground as evaluation frameworks mature

Anthropic Faces South Korean Privacy Investigation Over Data
Models

Anthropic Faces South Korean Privacy Investigation Over Data

South Korean regulator investigates Claude data exposure

BotPsy Runs Agentic AI on Phones With Sub-Billion Parameters
Models

BotPsy Runs Agentic AI on Phones With Sub-Billion Parameters

On-device multi-step reasoning finally scales to mobile, reshaping what local AI actually means

Claude Meetings: Anthropic Embedded in Google Enterprise AI
Models

Claude Meetings: Anthropic Embedded in Google Enterprise AI

Claude Meetings brings Anthropic's AI to Google Meet to rival Copilot Teams.

DeepSeek Inference Framework Hits 25K Stars in 25 Days
Models

DeepSeek Inference Framework Hits 25K Stars in 25 Days

DeepInfer's rapid adoption signals infrastructure maturity for locally-deployed open-source models.

Anthropic Advances Provable Inference Methods for AI Safety
Models

Anthropic Advances Provable Inference Methods for AI Safety

Formal verification research could enable auditable AI behavior

Local LLMs Hit a New Low: Running AI on 2015 Kindle Hardware
Models

Local LLMs Hit a New Low: Running AI on 2015 Kindle Hardware

Open-source optimization tools now enable inference on ultra-constrained devices

Anthropic's $16.3B Revenue IPO Filing Exposes Claude's Enterprise Dominance—And Rising Stakes
Models

Anthropic's $16.3B Revenue IPO Filing Exposes Claude's Enterprise Dominance—And Rising Stakes

Anthropic files S-1 with $16.3B revenue, signaling massive enterprise AI adoption ahead of public markets debut

NVIDIA Open-Sources Cosmos 3 World Models for Robotics and Simulation
Models

NVIDIA Open-Sources Cosmos 3 World Models for Robotics and Simulation

Open-source world models go mainstream with NVIDIA's Cosmos 3 release.

Anthropic Relaxes Game Development Ban, Signals Shift Toward Nuanced Safety Rules
Models

Anthropic Relaxes Game Development Ban, Signals Shift Toward Nuanced Safety Rules

Claude now permits game code generation, reversing blanket restriction

Open-Source Projects Challenge Proprietary AI Agents With MCP-Compatible Stacks
Models

Open-Source Projects Challenge Proprietary AI Agents With MCP-Compatible Stacks

Self-hosted alternatives gain momentum as enterprises prioritize cost control and data sovereignty

Mistral's Latest Frontier Model Arrives on HuggingFace, Challenging Llama's Open-Source Dominance
Models

Mistral's Latest Frontier Model Arrives on HuggingFace, Challenging Llama's Open-Source Dominance

Mistral Large 4 now available for self-hosting and local deployment

DeepSeek's New Open-Source Agent Framework Targets Production Autonomy—But Adoption Questions Remain
Models

DeepSeek's New Open-Source Agent Framework Targets Production Autonomy—But Adoption Questions Remain

Rapid GitHub growth masks gap between hype and real-world deployment

Malicious Claude Code Extensions Highlight Security Risks in Developer Tool Ecosystems
Models

Malicious Claude Code Extensions Highlight Security Risks in Developer Tool Ecosystems

Trojanized Claude Code plugins discovered with persistent SSH backdoors.

DeepSeek Harness Go and Android Ad-Cleaner Show Momentum for On-Device AI Agents Without Cloud Dependency
Models

DeepSeek Harness Go and Android Ad-Cleaner Show Momentum for On-Device AI Agents Without Cloud Dependency

New tools enable local LLM inference on mobile; ecosystem racing to escape cloud costs