GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
pyromind
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
pyromind
Model Tree
Base
Qwen/Qwen3.5-4B
Fine-tuned
this model
Input Modalities
Text
Image
Video
Output Modalities
Text
Supported Functionality
Dedicated Endpoints
| Item | Value |
|---|---|
| Base model | Qwen/Qwen3.5-4B |
| Init | PyroDash-4B-SFT |
| Stage | (3) GRPO · λ = 0.05 (quality-oriented) |
| Control token | <|llm_offload|> |
| SFT dataset | EasyHard-24K |
| GRPO dataset | BytedTsinghua-SIA/DAPO-Math-17k |
| Expert LLM (train/eval) | GLM-5.2-FP8 (frozen) |
| Precision | bfloat16 |
bash
git clone https://github.com/PyroMind-Dynamics/pyroDash.gitcd pyroDashpip install -r requirements.txt
evaluation/math_eval.sh)Edit placeholders in evaluation/math_eval.sh, then:
bash
bash evaluation/math_eval.sh
The script (1) starts a local vLLM server for the small model on port 8001, (2) runs math_eval.py, and (3) stops vLLM on exit.
| Variable / flag | Meaning | Example |
|---|---|---|
MODEL | Local merged model path (vLLM serve + tokenizer) | /path/to/your/merged_model |
--glm-base-url | OpenAI-compatible API for the large/relay model | http://your-glm-host:8000/v1 |
--glm-api-key | API key for that endpoint | your-glm-api-key |
Tokenizer must include the special token <|llm_offload|>.
| Method | Avg. Acc. (%) | LLM Token Ratio (%) | Avg. LLM Calls | Cost ($) |
|---|---|---|---|---|
| Qwen3.5-4B | 28.36 | 0.00 | 0.000 | 2.26 |
| Qwen3.5-4B (+SFT) | 46.25 | 0.00 | 0.000 | 1.32 |
| RouteLLM (~75% GLM-5.2-FP8) | 52.74 | 77.37 | 0.808 | 44.62 |
λ=0.05 is the quality operating point (Avg. Acc. 64.04%, above GLM-5.2-FP8 alone). For low cloud cost (~1.9% LLM tokens), use λ=0.6.
| Resource | Link |
|---|---|
| Project website | PyroMind-Dynamics.github.io/pyroDash |
| Code | github.com/PyroMind-Dynamics/pyroDash |
| SFT dataset (EasyHard-24K) | huggingface.co/datasets/pyromind/easyhard-24k |
| GRPO dataset (DAPO-Math-17k) | huggingface.co/datasets/BytedTsinghua-SIA/DAPO-Math-17k |
| Hugging Face org |
bibtex
@misc{pyrodash2026,title = {PyroDash: Cost-Efficient Token-Level Small-Large Model Collaborative Inference},author = {{PyroMind Dynamics}},year = {2026},note = {Preprint}}@misc{pyromind2026easyhard24k,title = {{EasyHard-24K} v0.02},author = {{PyroMind Dynamics}},year = {2026},howpublished = {\url{https://huggingface.co/datasets/pyromind/easyhard-24k}}}
GRPO training data: BytedTsinghua-SIA/DAPO-Math-17k.
Apache 2.0 (derived from Qwen3.5-4B).
--glm-model| Served model name on the GLM side |
your-glm-model |
--output-dir | Per-dataset JSON output directory | ./results_500 |
--datasets | Benchmarks (space-separated) | gsm8k minerva olympiad aime2024 aime2025 |
| GlimpRouter (τ=0.9) | 54.20 | 75.11 | 1.20 | 31.61 |
| PyroDash (λ=0.1) | 55.29 | 8.19 | 0.058 | 4.71 |
| PyroDash (λ=0.6) | 54.55 | 1.90 | 0.012 | 1.78 |
| PyroDash (λ=0.05) ← this | 64.04 | 95.34 | 0.975 | 39.29 |
| GLM-5.2-FP8 | 57.68 | 100.00 | 1.000 | 49.36 |