Discoveries / TurinTech

Faster code.
Faster engines.
Smaller models.

Artemis is a code optimization platform. Here is what it found when we pointed it at model compression, inference engines and customer code.

Run Artemis on your code
Explore the results
Get the models
changes merged into open-source projects
13
changes merged into open-source projects
open models made smaller, on Hugging Face
15
open models made smaller, on Hugging Face
smaller Gemma 4 E2B download
≈4×
smaller Gemma 4 E2B download

Results from our work with Artemis

Model compression

Open models made smaller with techniques Artemis found

Kev-27B27%
Gemma 4 E4B28%
Qwen3.5 9B32%

File size as a share of the original

See the models

Open source

Changes merged into projects like TensorRT-LLM and Whisper

TensorRT-LLM+40% throughput
NeMo-Speech.cpp1.74× faster whole-file diarization
whisper4.9× on the attention call
See the pull requests

Customer code

−75%

reported cost reduction for an investment firm’s optimized SQL queries

Before100%
After Artemis25%
See the SQL result

Latest highlights

The method behind the results

How we got these results

We give Artemis something to improve, a model or an engine's code, and a goal. It explores many methods and finds something new. TurinTech checks it, and we apply it with Artemis again and again. That is how we release many models quickly, and the same workflow produced our merged pull requests.

discovery · chat models

Input

Qwen3.5 9B

Qwen BF16 · 17.92 GB

ArtemisAsk

New discovery

Input: Qwen3.5 9B

Goal: A smaller file whose answers stay close to the original.

Smaller models appear here

The search and its timing are an illustration; a real Discovery runs for hours. Bars show each file as a share of the original download. Agreement comes from each card's own tests, which are small samples; some also informed which file was chosen.Browse the models

Step 1 of 5: Ask. We give Artemis a chat model and a goal.

Your next discovery

Use Artemis on the systems you run.

  1. 01

    Bring your workload

    Your code, a benchmark and a metric.

  2. 02

    Let Artemis search

    It tests candidate changes against your baseline.

  3. 03

    Review the evidence

    Keep the changes that win.

Run Artemis on your code
Explore the product