GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
Lego-X
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
Lego-X
Model Tree
Qwen/Qwen3.5-35B-A3B
Input Modalities
Output Modalities
Supported Functionality
The scaffold is part of the environment, not the policy. The same weights score very differently depending on which harness runs them, so training under a rewritten control flow optimizes for a deployment you never ship:
| Model | OpenHands SDK | Claude Code | OpenCode |
|---|---|---|---|
| Qwen3.5-35B-A3B (starting point) | 64.0 | 62.4 | 57.2 |
| Qwen3.6-35B-A3B (next-gen base) | 67.4 | 63.4 | 60.6 |
| KAT-Coder-V2.5-Dev (post-trained Qwen3.6) | 67.0 | 66.8 | 64.8 |
| Lego-RL-Qwen3.5-35B-A3B | 70.4 | 68.2 | 66.6 |
SWE-bench Verified (%), one shared protocol: temperature 0.7, 200 turns, 200K context.
Each Lego-RL column is a separate run trained in that harness from the same starting checkpoint, the same 2,699 tasks and the same 3 epochs. This repository is the Claude Code run (68.2). Across the three harnesses RL adds +6.4 / +5.8 / +9.4.
bash
vllm serve Lego-X/qwen3_5_35b_a3b_cc_200k_rl \--served-model-name vllm_model \--tensor-parallel-size 4 \--enable-expert-parallel \--max-model-len 262144 \--gpu-memory-utilization 0.9 \--enable-chunked-prefill --enable-prefix-caching \--dtype bfloat16 \--enable-auto-tool-choice \--tool-call-parser qwen3_coder \--host 0.0.0.0 --port 8000
[!IMPORTANT]
--tool-call-parser qwen3_coderis not optional. The model was rolled out and trained with this parser; serving it behindhermes(or any other parser) silently degrades tool-call formatting.
Point Claude Code at the vLLM endpoint as an Anthropic-compatible provider, and give it a repository workspace plus a 200-turn / 200K budget. The RL policy learned to spend turns; short turn caps systematically truncate the second half of its trajectories and cost most of the gain.
bash
git clone https://github.com/LegoX/Lego-RL.git && cd Lego-RLbash scripts/setup_env.shcp scripts/eval/_template.env scripts/eval/configs/my_eval.env # set MODEL_PATH, DATASET_PATH, kubeconfigbash scripts/eval/eval.sh scripts/eval/configs/my_eval.env
Sandboxed execution and verifier rewards come from Harbor; see the evaluation docs.
| Starting checkpoint | Qwen/Qwen3.5-35B-A3B (sparse MoE, 256 experts, 8 active) |
| Harness | Claude Code, unmodified — a thin adapter, not a fork |
| Released step | global_step_110 |
| Tasks | Lego-X/Lego-RL-2699 — 2,699 real repository issues, converted from GAIR/OpenSWE |
| Reward | each task's own test suite, run in a fresh sandbox: {0, 1}. No reward model, no patch similarity, no LLM judge |
The training set is disjoint from SWE-bench Verified at both the repository and the instance level.
Use it as an agent policy, not as a chat model: it was optimized inside a harness that hands it a repository, a shell, and file-editing tools.
| Lego-RL | harness-native RL for coding agents (this model) |
| SWE-Lego | the SFT recipe |
| SWE-Review | inference-time generate-review-revise |
| Terminal-Lego | trajectory-quality filtering |
Built on verl (trainer + rollout) and Harbor (sandboxed execution + verifier reward), with Claude Code as the harness.
bibtex
@misc{du2026legorlharnessnativereinforcementlearning,title={LEGO-RL: Harness-Native Reinforcement Learning for Coding Agents},author={Yiming Du and Yuxin Jiang and Tao Yuan and Jianbo Dai and Shaowei Wang and Jierun Chen and Chaofan Tao and Xianzhi Yu and Lifeng Shang and Kam-Fai Wong and Xiaohui Li and Haoli Bai},year={2026},eprint={2608.17393},archivePrefix={arXiv},primaryClass={cs.AI},url={https://arxiv.org/abs/2608.17393},}
| Algorithm | GSPO (sequence-level surrogate), group-relative advantage over G = 8 rollouts per task |
| Batch | 64 prompts × 8 responses = 512 trials/step; 3 epochs = 126 steps |
| Optimizer | lr 1e-6 constant, KL loss 1e-3 (low-var), clip [3e-4, 4e-4], rollout temperature 1.0 |
| Context | 200K, 200 turns |
| Backend | VeOmni FSDP, Ulysses SP = 8, R3 rollout routing replay; fully-async with partial rollout (staleness 1) |
| Hardware | 3 nodes × 8 GPUs (2 training + 1 rollout) |