whisper.cpp not run: our test log

We could not run whisper.cpp: no prebuilt Linux release was available for download and this machine has no make, g++ or cmake to build it.

By Agentic Weekly, operated independently from Texas, US · Published Oct 25, 2026 · Last updated Oct 26, 2026 (America/Chicago) · How we work

Tested by us on 2026-10-25. Local open-weight model, not the vendor's hosted models. A smoke test on a CPU-only box, not a benchmark.

Result at a glance

whisper.cpp test record
Field	Value
Result	Could not be run
Version tested	not run
Date and time	2026-10-25 (America/Chicago)
Task	Obtain a runnable whisper.cpp CLI and transcribe the same audio used for faster-whisper.
Model used	none (no language model involved)
Hardware	Linux x86_64 virtual machine (Debian 13), 8 vCPU (Intel Xeon, 1 thread per core), 15 GiB RAM, no GPU
Result	could not run

What we found

No transcription was run, so we publish no accuracy or speed figure for whisper.cpp. faster-whisper (separate test) covers the same job.

What we ran, step by step

Commands run, wall-clock time and exit code
Step	Wall time	Exit code	Notes
Look for a prebuilt Linux binary	n/a	1	No Linux x64 CLI asset was offered for us to download at the time we looked.
Build from source	n/a	n/a	We did not run the build: cmake, make and g++ are not installed on this machine, so we did not build it.

Exact commands (paths shortened)

# Look for a prebuilt Linux binary
check the GitHub releases for a Linux x64 CLI asset

# Build from source
cmake -B build && cmake --build build

Gotchas and things that broke

This is a limit of our machine, not a verdict on the project.

What we did not test

Everything.

Reproduce it

Versions change. These commands install whatever is latest on the day you run them unless a version is pinned, so expect different results later. The version we measured is in the table above.

git clone https://github.com/ggml-org/whisper.cpp && cd whisper.cpp && cmake -B build && cmake --build build -j

How to read this test

Every number on this page was measured on this one machine, which also ran other jobs, so timings are indicative and vary run to run. A pass means our small task worked once or as stated, a fail means it did not work as we ran it; neither says how the tool behaves with a vendor's frontier model, on a GPU, or at scale. Where a tool could not run without an account or a GPU, we say so instead of guessing.

See the testing rig for the hardware, the local model and the rules we follow, and the test hub for every other test.

ComfyUI: ComfyUI 0.42.0 installed in 17.42 s and its CPU-only server was ready in 30.1 s with 1014 node types, but we did not generate an image.
faster-whisper (speech to text): faster-whisper 1.2.1 transcribed 41 s of speech at 2.4 percent word error with the base.en model in float32 (deterministic), but the int8 setting gave 2 to 60 percent on identical audio across runs.
Piper (text to speech): Piper 1.8.0 synthesised 10 sentences (about 41 s of 22.05 kHz audio) on CPU in 1.4 to 1.5 s each including process start; 3.5 s uv install, 167 MB.

All test logs · How we test

FAQ

Why is there no whisper.cpp result?

It could not be run on our test machine without a build toolchain; we did not substitute a number.

Sources

whisper.cpp repository (fetched 2026-10-25)

Markdown version for agents: /tests/whisper-cpp.md