Where to run it
Prices per 1M tokens, read November 18, 2026.
| Host | Model id | Inputper 1M | Outputper 1M | Context | API formats | Access | |
|---|---|---|---|---|---|---|---|
| Featherless | huihui-ai/Huihui-Qwen3.8-27B-abliterated | $1.60 | $12.00 | 32K | Chat Completions | Plan or pay as you go | Visit |
Run it locally
About 17 GB at 4-bit, on one 24 GB GPU or a 32 GB Mac.
Ollama
ollama run huihui_ai/Qwen3.8-abliterated
GGUF weights
huggingface.co/huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF
Commands from the model card, the Hugging Face Ollama guide and the llama.cpp README. All local models.
Against the other models
Size
Total parameters, billions. Log scale. Larger is more capable
Context window
Tokens the model holds at once. Higher is better
Price
Blended USD per 1M tokens, lowest listed host. Lower is better
AbliteratedFine-tunedHatched: weights only, no API listed