GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
xupy21
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Run this model inference with full control and performance in your environment.
Model Details
Model Provider
xupy21
Model Tree
Base
Qwen/Qwen3-8B
Fine-tuned
this model
Input Modalities
Text
Output Modalities
Text
Supported Functionality
Dedicated Endpoints
Container
Across 5 long-horizon benchmarks (2 in-distribution agentic coding, 3 out-of-distribution), ContextRL improves over the standard GRPO baseline by +1.5 points on average, while improving every individual benchmark.
| Benchmark | Base | RL (GRPO) | ContextRL (Ours) |
|---|---|---|---|
| SWE-Bench Verified | 5.00 | 6.20 | 7.00 |
| SWE-Bench Lite | 2.70 | 2.70 | 4.00 |
| LiveCodeBench v6 | 44.6 | 46.3 | 47.4 |
| LongBench v2 (Overall) | 31.6 | 31.8 | 33.2 |
| LongBench v2 (Long) | 27.8 | 26.9 | 29.6 |
| NIAH | 98.8 | 98.5 | 99.0 |
Metrics: SWE-Bench Verified/Lite resolve rate (%), LiveCodeBench v6 solve rate (%), LongBench v2 accuracy (%), NIAH mean recall (%). On the long-context tasks (LongBench v2, NIAH) where standard outcome-based GRPO struggles or regresses, ContextRL surpasses both the base model and the RL baseline, demonstrating strong out-of-distribution generalization.
This model follows the same interface as Qwen3-8B and can be loaded with transformers.
Training and evaluation code, data construction pipelines, and detailed configurations are
available in the repository:
👉 https://github.com/xupy2003/ContextAwareRL
Please refer to the repo's README for environment setup, inference scripts, and
reproduction instructions.