unconst/Affine-5czsc2fc98-r225-reinforce

TEXT GENERATIONPricing:Input $0.4 / Cached $0.07 / Output $4Concurrent Unit Cost:3Model Size:35.1BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 14, 2026Architecture:Transformer Featherless Exclusive Cold

Affine-5czsc2fc98-r225-reinforce is a 35.1 billion parameter model developed by unconst, created by LoRA-merging from `kevin954/Affine-5dfqbbh8ev-sft`. This model is noted as a private TTL insurance checkpoint, indicating its development for specific, internal reinforcement learning applications rather than general public use. Its primary characteristic is its origin as a merged checkpoint, suggesting a focus on refining or specializing an existing base model for particular tasks.

Loading preview...

Model Overview

The unconst/Affine-5czsc2fc98-r225-reinforce model is a 35.1 billion parameter language model. It was created through a LoRA-merge process, specifically derived from the kevin954/Affine-5dfqbbh8ev-sft checkpoint.

Key Characteristics

  • Parameter Count: 35.1 billion parameters, indicating a substantial capacity for complex language understanding and generation tasks.
  • Context Length: Supports a context length of 32768 tokens, allowing it to process and generate longer sequences of text.
  • Development Status: Described as a "private TTL insurance" checkpoint, suggesting it is an internal development or backup version, not intended for general release or public consumption at this stage.
  • Origin: The model is a merged checkpoint, implying it incorporates fine-tuning or adaptations from a prior supervised fine-tuning (SFT) model, likely for specific performance enhancements.

Intended Use

This model is explicitly noted as "not a submission until Stage-5 gate clears," indicating it is in an early, private development phase. It is likely being developed for specialized internal applications, possibly involving reinforcement learning, given the "-reinforce" in its name. Developers should note its private and developmental status, as it is not yet positioned for broad public use cases.