Currently listed through these providers:
Model details
Gemma 4 31B Queen
Gemma 4 31B Queen sits inside the broader Gemma family as an instruction-tuned community build rooted in a Gemma-4-Queen architecture, with public weights distributed through Hugging Face under the aifeifei798 namespace alongside companion GGUF quantizations. The supplied Hugging Face page frames the model around an unusual focal point: logic density and physical-world modeling rather than raw parameter scaling, positioning the 31B size as a deliberate trade-off. The presence of a QAT-quantized unquantized-format variant and a declared transformers 5.5.0 dependency signal that the author is targeting practical, deployable efficiency for downstream users, including local and edge-style use.
According to the hosted provider description, the model is oriented toward structured deductive tasks such as causal-chain analysis, sealed-room reasoning puzzles, and persona-locked multi-turn dialogues where resisting instruction drift matters. The Hugging Face README reinforces that profile with a worked "Case File" demonstrating timestamped-event reconstruction, physical-constraint reasoning, and careful integration of contradictory evidence. That combination of instruction rigidity, scenario-aware deduction, and creative role-play personas makes the model a reasonable fit for projects that need a mid-sized open-weights engine for complex reasoning, technical writing, or character-driven interactive fiction rather than sheer breadth-of-knowledge workloads.
Quick Info
Powered by- Provider
- NanoGPT
- Model key
- Gemma-4-31B-Queen
- Release date
- May 1, 2026
- Last updated
- May 1, 2026
- Input modalities
- Output modalities
- Capabilities
Cost
- Input token cost
- $0.306
- Output token cost
- $0.306
Limits
- Input tokens
- 262,144 tokens
- Output tokens
- 16,384 tokens
- Context window
- 262,144 tokens
Latest news about Gemma 4 31B Queen
No articles yet. Fetch the latest news to show it here.