← Founder Notes
Archive

The lab whose moat was a closed model just gave away the harness. deepseek open-sourced dsh this…

Yethikrishna ROriginal on Threads

the lab whose moat was a closed model just gave away the harness. deepseek open-sourced dsh this week and hit 150k stars in days on an everything-as-a-plugin build.

you don't win on weights anymore, you win on who runs your orchestration in prod.

Context

DeepSeek's harness page says DeepSeek Harness is in public preview worldwide and open source, with composable plugins, and that everything is a plugin; a search result dated it 12 August 2026. The repository is MIT licensed and showed 212,481 stars when read. GitTrend, fetched 4 October 2026, lists 100k stars on 15 August and 150k on 17 August 2026.

How it compares

This week does not match the first-party date of 12 August 2026 for the release. In days fits the August star crossings, not the week of 8 September. The closed-model premise is not established: no inspected evidence either way about the lab's prior moat. You do not win on weights and orchestration being the moat are the author's opinion. Orchestration products from GitHub (HydraFusion) and NVIDIA (Switchyard) exist, but GitHub's results are vendor-run offline tests.

Related work

Watch next

  • Which models the harness drives, and adoption data beyond stars.

Sources

  1. DeepSeek Harness (DeepSeek)deepseek.com
  2. deepseek-ai/deepseek-harness (GitHub)github.com

Provenance

The note above is reproduced unedited from the original post, first published on Threads on 8 September 2026 at 17:11 IST. Sources are the papers and datasets the note draws on.

View the original post
Embed this note
<iframe src="https://founder.myndlabs.tech/notes/embed/the-lab-whose-moat-was-a-closed-model-DdBmhLmjH5I" width="480" height="420" style="border:0;max-width:100%" loading="lazy" title="The lab whose moat was a closed model just gave away the harness. deepseek open-sourced dsh this…"></iframe>

More notes