Valen

tom@tomrochette.com (Tom Rochette) — Tue, 13 Oct 2026 00:00:00 +0000

Valen is an Apache-2.0 open-weights multimodal decision model in the Jev family: text, images, and video in, candidate probabilities out, shipped as 0.8B, 2B, and 4B checkpoints with the training recipe, general dataset, and evaluation sets published beside them.

Its angle is reproducibility: it is the first Jev-family member that ships the whole pipeline, training code (SFT plus experimental RLCD), a 100k general dataset, the VisualDecisionBench evaluation, and ONNX and GGUF exports, so the decision layer is something you retrain rather than download and trust.

What it is
#

A training repository plus model family from an independent team, explicitly inspired by the Jev System One model lineage, with vision added: the models read screenshots, documents, and video frames, and emit decision probabilities for routing, gating, and classification calls. The Hugging Face org carries the three model cards, the training and evaluation datasets, and a live demo Space; ONNX and GGUF releases landed the week of this check, so the models run on modest local hardware and in runtimes beyond PyTorch. The README reports results on VisualDecisionBench and a game-evaluation set, with the Jev lineage anchor (the typesafe.ai System One post) cited directly.

Status
#

Very young and actively shipping: 774 stars, 70 forks, and 0 open issues as of 2026-10-13, created 2026-09-23 and pushed the day of this check, with the weights, datasets, demo, and export builds all public. The footprint is the thinnest in its family: a single 2-point Hacker News post and no third-party coverage found, so every quality claim is currently the team’s own. The 0-open-issue count reflects age, not maturity.

Strengths
#

Full-pipeline openness: training code, data, benchmarks, and exported weights are all published, which is the preconditions for actually verifying a decision model.
Vision extends the Jev pattern to the cases that need it (screenshots, documents, video frames) at 0.8B to 4B scale.
The experimental RLCD stage beyond SFT is published rather than hidden, warts included.
Export builds (ONNX, GGUF) make deployment boring: the models run in standard runtimes on small hardware.
Apache-2.0 with no API dependency, so the decision layer cannot be revoked by a vendor.

Cautions
#

Every benchmark number is self-reported on self-built benchmarks (VisualDecisionBench, Valen-Eval-Game), with no independent evaluation found.
Three weeks old at verification with one maintainer team and a single 2-point HN post; treat the project as a pre-adoption artifact.
RLCD is labeled experimental by the authors themselves, so the training recipe’s stability is unproven.
The Jev family’s known weakness (calibration drift outside the training distribution) applies with no published out-of-distribution numbers.
Probabilities out means no schema enforcement: the caller owns validation, the opposite of this category’s structured-output columns.

Pricing
#

Free and open source under Apache-2.0, weights and all. No hosted tier; the cost is your own inference compute, which at 0.8B to 4B is small.

Compared to
#

H2O-Lightning-4B: the family’s established open-weights member with a longer track record; Valen adds vision and the full training pipeline.
Strands Decider 2B: AWS’s decision model with the better-published training discipline; Valen matches the openness and adds multimodality at a smaller scale.
Jev: the original closed System One model whose pattern Valen replicates openly.

Bottom line
#

Recommended for engineers who want a small multimodal decision model they can audit, retrain, and run locally, and who will validate it on their own data. Not for production routing without independent evaluation, or for anyone needing schema-enforced tool calls from the decision layer itself.

Changes
#

2026-10-13 - Created.

References
#

https://github.com/Liuziyu77/Valen - repository, Apache-2.0 license, stars, push activity, and the training-and-eval structure as of 2026-10-13
https://huggingface.co/Valen-Team - the Hugging Face org holding the three model cards and datasets
https://huggingface.co/Valen-Team/Valen-4B - the flagship model card with usage and reported results
https://huggingface.co/datasets/Valen-Team/VisualDecisionBench - the published evaluation set behind the benchmark claims
https://huggingface.co/spaces/yuhangzang/Valen-Preview-0923 - the live demo Space
https://typesafe.ai/blog/introducing-system-one-models-and-jev - the System One and Jev lineage post the README cites
https://news.ycombinator.com/item?id=49838068 - the single 2-point HN post, the recorded thin community footprint

decision-model on tomrochette.com

Valen

What it is
#

Status
#

Strengths
#

Cautions
#

Pricing
#

Compared to
#

Bottom line
#

Changes
#

See also
#

References
#

decision-model on tomrochette.com

Valen

What it is #

Status #

Strengths #

Cautions #

Pricing #

Compared to #

Bottom line #

Changes #

See also #

References #

What it is
#

Status
#

Strengths
#

Cautions
#

Pricing
#

Compared to
#

Bottom line
#

Changes
#

See also
#

References
#