Topic
Tool use, agent systems, evaluation, deployment, and the model releases that shape them.
AI Agents
DeepSeek's experimental V4 Flash Vision model adds image understanding to the API at V4 Flash pricing, with Chat Completions, Messages, Responses and Files API support.
AI Agents
Muse Glimmer is Meta’s 30B multimodal open model for local agent work. Its unusually lean KV cache is the part worth watching when a task runs long.
AI Agents
A practical way to tell whether an agent needs a safer runtime, a better feedback loop, or explicit workflow control.
AI Agents
For readers outside China, the useful signal is not a leaderboard position. It is how a workplace agent combines model routing, tool execution and a familiar office workflow into one product surface.
AI Agents
Memory, repository instructions and tool permissions are not housekeeping. They decide what an agent can know, what it may do, and whether its work remains reviewable after the conversation ends.
Industry reporting
A reporting-based review that separates news coverage from an official DeepSeek announcement or confirmed meeting transcript.
AI Agents
The connector makes Anthropic’s usage data easier to query. It does not turn that dataset into a complete view of the labor market.
Industry reporting
A clearly attributed reading of reporting on Baidu’s internal OKRs and what it may indicate about AI commercialization priorities.
Model releases
A source-first reading of Moonshot’s Kimi K3 announcement, documentation, and the questions that still need the technical report.
Model releases
A concise source list for OpenAI’s July GPT-5.6 release and the product-specific availability details readers should verify.
Model releases
What Alibaba has released, what its architecture claims mean in practice, and which details belong to provider-reported evaluation.