SYLIN

Hugging Face

Maker · 7 entries

Everything from Hugging Face

NameWhat to knowPrice or memoryRanks
Qwen3.8-27B opencode-adaptedModelOpenCode adapter moved Qwen3.8-27B from 35.3% to 51.0% on the hard SWE-bench Pro subset with thinking off
Qwen3.8-27B pi-adaptedModelCorrected Pi adapter result on the 51 hardest SWE-bench Pro tasks is 20 to 33 solves (39.2% to 64.7%), after a tunnel outage hit nine base runs
Hugging Face Inference ProvidersApp or serviceNow runs on OpenRouter with the same endpoint and token, one endpoint for open models
Llama desktop appsApp or serviceunproven Only Hugging Face's announcement exists so far; no one has tested these apps.
TransformersFrameworkAdds a shared model.decide() interface for decision models
Carbon-XModel14B DNA understanding and generation model from Qwen3, trained on 300B DNA tokens plus RL; under 30 GB, Apache 2
Transformers.jsFramework

Plans

What each one gives you.
Hugging Face PRO $9/month$2 inference credits a month, 8x ZeroGPU quota, 10x private storageHugging Face Inference Providers