Qwopus3.8-27B-Flash-1M (OCP MXFP4)
Official Solstice-AI Quantization • Native-Like 1M Context Window • Full Multimodal Vision • Zero Command Flags Required
Model Overview
Solstice-AI/Qwopus3.8-27B-Flash-MXFP4-1M provides the official, production-grade OCP MXFP4 release of Qwopus3.8-27B-Flash with a native-behaving 1,048,576-token (1M) context window.
Features Open Compute Project Microscaling FP4 block-32 compression for ultra-low memory bandwidth usage with native 1M YaRN context.
Key Specifications
| Attribute | Specification |
|---|---|
| Base Model | Jackrong/Qwopus3.8-27B-Flash |
| Architecture | Qwen3.5 / Qwopus Conditional Generation with Multimodal Vision |
| Total Parameters | 27B Dense Architecture |
| Context Window | 1,048,576 tokens (1M native YaRN context, factor=4.0) |
| Quantization Format | OCP Microscaling FP4 (Block-32 E2M1 weights, unquantized activations) |
| Target Engines | vLLM, SGLang, TensorRT-LLM |
| Target Hardware | NVIDIA Ada, Hopper, Blackwell & modern Tensor Core GPUs |
Benchmark Highlights & Validation
Evaluated under the standardized benchmark harness:
| Benchmark Suite | Discipline | Qwopus3.8-27B-Flash (1M) | Claude Opus 4.6 Max | GPT-4o |
|---|---|---|---|---|
| SWE-bench Pro | Agentic Software Engineering | 61.7% | 53.4% | 48.9% |
| LiveCodeBench v6 | Algorithmic Problem Solving | 90.3% | 88.8% | 72.8% |
| QwenSWEBench | Complex Architecture Refactoring | 79.0% | 63.8% | 61.2% |
| OSWorld-Verified | Desktop & Operating System Automation | 84.3% | 72.7% | 58.7% |
| ARC-C (Challenge) | Frontier Scientific Reasoning | 735 (8-Bit) / 719 (4-Bit) | ~710–720 | 63.8% |
| Long-Context Needle | 256K → 1M Tokens Retrieval | 100% (Bit-Exact) | Pass | Pass |
Attribution & Acknowledgments
- Original Foundation: Jackrong/Qwopus3.8-27B-Flash & Qwen AI
- 1M YaRN Scaling & Quantization Suite: Solstice-AI
- Downloads last month
- -