gradients-io-tournaments/swe-base-qwen3-8b-continuous

TEXT GENERATIONPricing:Input $0.468 / Output $1.82Concurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 27, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

The gradients-io-tournaments/swe-base-qwen3-8b-continuous model is an 8 billion parameter causal language model based on the Qwen3 architecture, specifically derived from Qwen/Qwen3-8B-Base. It was a winner in a Gradients tournament and serves as the target base model for the SWE environment tournament. This model retains a tool-calling chat template, making it suitable for tasks requiring structured interaction and function calling.

Loading preview...

Model Overview

gradients-io-tournaments/swe-base-qwen3-8b-continuous is an 8 billion parameter causal language model built on the Qwen3 architecture (Qwen3ForCausalLM). This model originated as a winning entry in a Gradients tournament and is specifically designated as the target base model for the swe_infinite environment tournament.

Key Characteristics

  • Architecture: Based on Qwen3ForCausalLM, seeded from Qwen/Qwen3-8B-Base.
  • Lineage: It represents the final chunk (train_index 8 winner) of a retired continuous-SFT lineage focused on qwen models.
  • Purpose: Primarily intended as a base model for the SWE (Software Engineering) environment tournament.
  • Chat Template: Retains a tool-calling chat template (chat_template.jinja), suggesting capabilities for structured interactions and function calling.
  • Integrity: The weights are identical to its source repository, with only naming and packaging adjustments made for this release.

Use Cases

This model is particularly relevant for:

  • Software Engineering Tasks: Its role as a base model for the SWE environment tournament indicates potential for code-related tasks, problem-solving in software contexts, and automated development workflows.
  • Tool-Calling Applications: The preserved tool-calling chat template makes it suitable for applications requiring the model to interact with external tools or APIs by generating structured calls.
  • Research and Development: As a tournament-winning model, it can serve as a strong baseline for further fine-tuning or research in language model capabilities, especially within the Qwen3 family.

Finetunes
1 models