maylore-ai/Temper-1-0.5B

TEXT GENERATIONPricing:Input $0.04 / Cached $0.008 / Output $0.08Concurrent Unit Cost:1Model Size:0.5BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Oct 5, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

Temper-1-0.5B is a 0.5 billion parameter language model developed by maylore-ai, fine-tuned from Qwen2.5-Coder-0.5B-Instruct with a 32768 token context length. It is specifically optimized for generating Rust code, demonstrating superior pass@1 rates on Rust coding tasks compared to models up to three times its size. This model is ideal for developers requiring efficient and accurate Rust code generation, particularly for small to medium complexity problems.

Loading preview...

Overview

Temper-1-0.5B is a compact 0.5 billion parameter model developed by maylore-ai, specifically fine-tuned for generating Rust code. It is based on the Qwen2.5-Coder-0.5B-Instruct architecture and has been trained on 10,000 Rust examples generated by DeepSeek-V4-Flash, which were then validated by compiler and tests. Further refinement was achieved through reinforcement learning on its own generated answers.

Key Capabilities

  • Specialized Rust Code Generation: Outperforms models up to 1.5B parameters in pass@1 scores on various Rust coding benchmarks, including internal trivial, easy, and medium sets, as well as HumanEval-RS and MBPP-RS.
  • Efficiency: Achieves strong performance in Rust code generation with a significantly smaller parameter count, making it suitable for resource-constrained environments.
  • Chat-based Interaction: Designed to accept chat messages as input for code generation tasks.

Performance Highlights

Temper-1-0.5B shows notable pass@1 results:

  • 81.8% on trivial Rust tasks
  • 59.0% on easy Rust tasks
  • 18.0% on medium Rust tasks
  • 34.7% on HumanEval-RS
  • 38.8% on MBPP-RS

Good for

  • Developers needing a highly efficient and accurate model for generating Rust code.
  • Applications requiring quick first-attempt solutions for Rust programming challenges.
  • Integrating Rust code generation into chat-based interfaces.

Limitations

  • Primarily trained for Rust code generation; performance on other languages or non-Rust related queries may be poor.
  • Not designed for fill-in-the-middle code completion.
  • Limited training data for async code and web frameworks in Rust.

Finetunes
2 models