Tenstorrent just opened its model compiler to everyone. tt-forge, the mlir-based stack built on the…
tenstorrent just opened its model compiler to everyone. tt-forge, the mlir-based stack built on the company's ai software, hit public beta on september 3, compiling pytorch, jax and onnx models for its hardware.
the compiler caught up to the chip's ambitions.
Context
Tenstorrent's TT-Forge page says TT-Forge is Tenstorrent's MLIR-based compiler stack for compiling, optimizing, debugging and extending models, and that it is now in public beta. It is designed to connect with OpenXLA, MLIR, ONNX, TVM, PyTorch and TensorFlow. TT-XLA, built on PJRT and StableHLO, handles JAX and PyTorch with multi-chip execution, and TT-Forge-ONNX is the ONNX frontend. The stack lowers to TT-NN and TT-Metalium.
The page is undated, so September 3 as the beta date is unverified. The page also lists TensorFlow, which the note omits, and that omission is not an error. The license and exact hardware and model coverage were not read. Opened to everyone is the author's framing, and open-source flexibility is the page's own wording. Public beta status does not by itself establish which hardware is supported or which models compile. The compiler caught up to the chip's ambitions is the author's take.
Watch next
- A dated release or blog post for the beta, model coverage lists and the license.
Sources
- Tenstorrent: TT-Forgetenstorrent.com
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 23 September 2026 at 01:48 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →