GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
gatesushi
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
gatesushi
Model Tree
Input Modalities
Output Modalities
Supported Functionality
Based on nightmedia's Qwen3.6-27B-Architect-Polaris2-Fable-B-F451
This quantization was performed using omlx (0.5.4 dev). It utilizes mtp and oQ quantization.
The Model appears to perform better in benchmarks and has seemed effective in chats.
Here are the benchmarks executed in omlx. These are 100 tests with the think set to off.
Model
Benchmark Accuracy Time(s)--------------------------------------------------------------MMLU 90.0% 145.3MMLU_PRO 73.0% 65.8HELLASWAG 94.0% 64.6TRUTHFULQA 91.0% 44.6ARC_CHALLENGE 91.0% 41.2WINOGRANDE 85.0% 39.1GSM8K 96.0% 308.7MATHQA 59.0% 44.7Model: Qwen3.6-27B-Architect-Polaris2-Fable-B-F451-mxfp4-mlxBenchmark Accuracy Time(s)--------------------------------------------------------------MMLU 88.0% 116.4MMLU_PRO 77.0% 59HELLASWAG 93.0% 62.2TRUTHFULQA 90.0% 43.1ARC_CHALLENGE 92.0% 39.5WINOGRANDE 82.0% 37.4GSM8K 96.0% 338.4MATHQA 52.0% 44.8Model: Qwen3.6-27B-Miraculix-mxfp8-mlxBenchmark Accuracy Time(s)--------------------------------------------------------------MMLU 82.0% 127.1MMLU_PRO 73.0% 63.4HELLASWAG 93.0% 65.5TRUTHFULQA 91.0% 48.3ARC_CHALLENGE 92.0% 45.7WINOGRANDE 83.0% 43.9GSM8K 96.0% 344.7MATHQA 54.0% 54Model: Qwen3.6-27B-Fable-Fusion-711-BigBubba-717-mxfp8-mlxBenchmark Accuracy Time(s)--------------------------------------------------------------MMLU 89.0% 158.2MMLU_PRO 75.0% 74.2HELLASWAG 93.0% 72.9TRUTHFULQA 93.0% 52.7ARC_CHALLENGE 92.0% 49.1WINOGRANDE 83.0% 47GSM8K 97.0% 454.8MATHQA 55.0% 60.9Model: Qwen3.6-27B-ElCHUD-mxfp8-mlxBenchmark Accuracy Time(s)--------------------------------------------------------------MMLU 88.0% 171MMLU_PRO 75.0% 78HELLASWAG 93.0% 74.2TRUTHFULQA 92.0% 53ARC_CHALLENGE 92.0% 49.5WINOGRANDE 84.0% 47GSM8K 96.0% 447.8MATHQA 55.0% 55.1Model: Qwen3.6-27B-Akka-mxfp8-mlxBenchmark Accuracy Time(s)--------------------------------------------------------------MMLU 84.0% 132.5MMLU_PRO 72.0% 75.1HELLASWAG 92.0% 74.4TRUTHFULQA 92.0% 53.4ARC_CHALLENGE 91.0% 50.9WINOGRANDE 83.0% 49.9GSM8K 97.0% 425.7MATHQA 53.0% 57