Outlier-Ai/Outlier-10B-V2

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:7.6BQuant:FP8Ctx Length:32kPublished:Apr 7, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Warm

Outlier-Ai/Outlier-10B-V2 is a research artifact from an earlier Outlier lineage, preserved for reproducibility and historical reference. This model predates the current Outlier lineup and is not recommended for production use. It may contain experimental adapters, overlays, or quantization variants that were not integrated into later shipping products. Developers seeking current Outlier models should refer to the Outlier Nano 4B, Core 27B, Vision 35B-A3B, DeepSeek-R1-Distill-Qwen-7B, or Qwen3-Coder-30B-A3B.

Loading preview...

Outlier-10B-V2: A Historical Research Artifact

Outlier-10B-V2 represents an earlier development stage within the Outlier lineage. This model is explicitly noted as superseded and is preserved primarily for reproducibility and historical reference, rather than as a recommendation for current production environments.

Key Characteristics:

  • Superseded Status: This version has been replaced by more recent and optimized Outlier models.
  • Experimental Nature: It may incorporate experimental adapters, overlays, or quantization techniques that did not advance to the shipping product line.
  • Not for Production: Users are advised against deploying this specific model for new applications.

What Replaced It:

For current development and production needs, Outlier-Ai recommends their updated lineup, which includes:

  • Outlier Nano 4B: The current entry-tier model.
  • Outlier Core 27B: The current default tier model.
  • Outlier Vision 35B-A3B: A multimodal tier model.
  • DeepSeek-R1-Distill-Qwen-7B: A popular model optimized for reasoning tasks.
  • Qwen3-Coder-30B-A3B: A popular model designed for coding tasks.

Developers should visit outlier.host for the latest verified benchmarks and model downloads.