This product was not featured by Product Hunt yet. It will not be visible on their landing page and won't be ranked (cannot win product of the day regardless of upvotes).
Product upvotes vs the next 3
Waiting for data. Loading
Product comments vs the next 3
Waiting for data. Loading
Product upvote speed vs the next 3
Waiting for data. Loading
Product upvotes and comments
Waiting for data. Loading
Product vs the next 3
Loading
KALYPSO
Open 14B coding model, on a dataset you can inspect
KALYPSO-v1.1L is an open 14B coding model (built on Qwen2.5-Coder-14B), trained on Kraken: 18,049 agentic + coding examples we curated and decontaminated against HumanEval & MBPP. We shipped the weights, the full dataset, and the decontamination pipeline. Inspect everything.
Hi Product Hunt 👋
We just shipped GENOMA Labs' first open release: KALYPSO-v1.1L, plus
the Kraken dataset it was trained on.
What it is: a 14B coding model on Qwen2.5-Coder-14B, trained on Kraken -
18,049 agentic + coding examples we curated and, crucially, decontaminated.
What it does:
- Generates Python, C++, and SQL
- Trained on agentic, multi-step tool-use trajectories, not just single-shot
snippets
- Runs locally on a single GPU
The numbers (EvalPlus, greedy):
- HumanEval: 87.8%
- MBPP: 84.7%
Why we think it's worth your time:
Most "open" releases hand you weights and a leaderboard number, and quietly
skip the part where benchmark problems leaked into the training set. We did
the opposite. Every Kraken record went through a 13-gram decontamination pass
against HumanEval & MBPP, and we published the removal report.
So when we say 87.8%, it's a number you can verify, on data you can inspect.
You don't just get a model. You get:
- The model: KALYPSO-v1.1L
- The full dataset: Kraken (18K records)
- The decontamination pipeline + the report
Pull it. Run it. Try to break it. Build on it. Then tell us what you find.
Built with Qwen. Trained on NVIDIA Nemotron data.
About KALYPSO on Product Hunt
“Open 14B coding model, on a dataset you can inspect”
KALYPSO was submitted on Product Hunt and earned 0 upvotes and 1 comments, placing #122 on the daily leaderboard. KALYPSO-v1.1L is an open 14B coding model (built on Qwen2.5-Coder-14B), trained on Kraken: 18,049 agentic + coding examples we curated and decontaminated against HumanEval & MBPP. We shipped the weights, the full dataset, and the decontamination pipeline. Inspect everything.
On the analytics side, KALYPSO competes within Open Source, Developer Tools and Artificial Intelligence — topics that collectively have 1.1M followers on Product Hunt. The dashboard above tracks how KALYPSO performed against the three products that launched closest to it on the same day.
Who hunted KALYPSO?
KALYPSO was hunted by genoma labs. A “hunter” on Product Hunt is the community member who submits a product to the platform — uploading the images, the link, and tagging the makers behind it. Hunters typically write the first comment explaining why a product is worth attention, and their followers are notified the moment they post. Around 79% of featured launches on Product Hunt are self-hunted by their makers, but a well-known hunter still acts as a signal of quality to the rest of the community. See the full all-time top hunters leaderboard to discover who is shaping the Product Hunt ecosystem.
For a complete overview of KALYPSO including community comment highlights and product details, visit the product overview.
genoma labs