lev

Released
2026-09-24
Last refreshed
2026-10-05
Status
Researched 1d ago
Open sourceCommercial use: permittedAgentsClassificationJSON / Tool useAgent

lev is available now for agents, classification, and json / tool use with open-source; evaluate it while provider pricing coverage matures.

Interfaze releases · 1 in the last 12 monthsChangelog →
Specifications
Family
lev
Released
2026-09-24
Parameters
4B
Openness
Open source
License
Apache 2.0OSI-approvedCommercial use: permitted
Weights
Available
Code
Unknown
Training
Fine-tuned
Pricing

No tracked provider token pricing is available yet.

About

lev is Interfaze's open decision model, a rank-32 LoRA adapter with a candidate-path head on Qwen3.5-4B. Given a state and a set of yes/no ('noul'), choice, and score questions, it reads each answer from logits it already computed, in one forward pass with zero output tokens, and returns calibrated probabilities over exactly the options supplied. Label-token readout handles yes/no, 2–10-level scores, and choice lists of up to several hundred options; options are read in two orders and averaged, with per-bucket temperatures.

Provider price ladder

No tracked provider token pricing is available for this model yet.

Capabilities

Structured Outputs

Benchmark peer barsfor Agents

No task-mapped benchmark peers are available for this model yet.

Benchmark scores(1)

Scores are benchmark-specific and are direction-aware: the same numeric gap can mean very different outcomes across suites. Use the leaderboard context and this model's provider route to decide whether the winning margin is meaningful for your workload.
BenchmarkScoreVersionEvaluationSource
JevBench42.2v1.6.0Observed 2026-10-04—Source

Migration checks

No linked migration route is available for this model yet.

API versions

interfaze-ai/levlevLev