- BTL-3 is a 27B open-weight agent model designed for agentic coding, structural tool use, repository work, failure recovery, and long multi-turn execution.
- It achieves high scores on benchmarks: 95.1% HumanEval pass@1, 88.5% BFCL v4 AST, and 88.1% LiveCodeBench v6.
- A Compact edition (8.39 GB) packages the full model into a single file with 92.2% behavior retention, enabling native local inference.
- The model is a PEFT LoRA adapter (rank 32, alpha 64) based on Qwen3.6-27B, with a 262,144-token context window and Apache-2.0 license.
- It can be used with Transformers (via PEFT) or vLLM, with recommended non-thinking mode for stability.
- Intended use cases include coding, debugging, test-driven repair, structured function calling, and long multi-turn tasks requiring verification and recovery.
- Operational guidance recommends sandboxing generated code and requiring confirmation for high-impact actions.