Topic

AI

Everything we publish on AI: the models we run on our own machines, what the tools send home, the benchmarks we measure ourselves and the claims we check against the code. Start with the investigations, then the guides.

Three publishers, one schedule: the quantisation policy we called a choice is a line of llama.cppNote
Research

Three publishers, one schedule: the quantisation policy we called a choice is a line of llama.cpp

A llama.cpp quantisation rule decides which layers get more bits. One line predicts all 20 promoted blocks in three publishers'…

7 min
The speculative decoding head ships two different ways, and neither one is in the model you downloadedNote
Research

The speculative decoding head ships two different ways, and neither one is in the model you downloaded

An MTP draft model GGUF ships either as an extra block inside the model or as a separate 49-tensor file.…

7 min
Experts really do go missing from MoE models, and the builds that remove them say so in the metadataNote
Research

Experts really do go missing from MoE models, and the builds that remove them say so in the metadata

REAP pruned MoE builds remove a quarter of the experts and declare it. We read four GGUF tensor tables: 48…

7 min
Two GGUF builds of the same model differ by a whole block, and it is the speculative decoding headNote
Research

Two GGUF builds of the same model differ by a whole block, and it is the speculative decoding head

One Qwen3.5 GGUF build ships 41 blocks and another ships 40. The extra block is the MTP head for speculative…

8 min
Hugging Face ships a telemetry function nothing calls, and a header that reports your PyTorch versionNote
Research

Hugging Face ships a telemetry function nothing calls, and a header that reports your PyTorch version

Hugging Face telemetry has an opt-out env var and a send function the library never calls. The data trail is…

7 min
Which local AI tools give your machine a permanent name, and which only look like they doNote
Research

Which local AI tools give your machine a permanent name, and which only look like they do

A local AI machine ID turns anonymous requests into a profile. We swept five installed tools: a naive search says…

8 min
Every local AI app ships a crash reporter, and none of the three we checked turns it onNote
Research

Every local AI app ships a crash reporter, and none of the three we checked turns it on

A crash reporter in a local AI app looks alarming in Activity Monitor. We checked three, and the crashpad process…

7 min
OpenCode contacts Sentry before you type anything, and two other findings that were not realNote
Research

OpenCode contacts Sentry before you type anything, and two other findings that were not real

OpenCode telemetry, measured on a live launch. A production Sentry DSN is baked into the app and it connects to…

8 min
The smallest Kimi K3 you can download is 589 GB, and two builds of it disagree by 58Note
Research

The smallest Kimi K3 you can download is 589 GB, and two builds of it disagree by 58

Running Kimi K3 local is not close to possible on a Mac. We summed every shard: the smallest complete build…

7 min
LM Studio does not track you, and it routes every model search through its own serversNote
Research

LM Studio does not track you, and it routes every model search through its own servers

LM Studio privacy, tested on a live install. No telemetry keys and no outbound connections at rest, but model search…

8 min