Step-3.5-Flash-Base: StepFun releases lightweight text generation model
StepFun has released Step-3.5-Flash-Base, a text generation model available on Hugging Face under Apache 2.0 license. The model is part of the Step 3.5 series and focuses on efficient inference.
Step-3.5-Flash-Base at a glance
How Step-3.5-Flash-Base ranks now
| Benchmark | Score | Rank | Source |
|---|---|---|---|
| MMLU-Pro | 62.3% | 67 of 73 | Hugging Face model card ↗ |
Rank is against every model in our database with a sourced score on that benchmark, each counted once with its best score.
StepFun Releases Step-3.5-Flash-Base
StepFun has released Step-3.5-Flash-Base, a text generation model designed for efficient inference. The model is available on Hugging Face as an open-source release under the Apache 2.0 license.
Model Details
The Step-3.5-Flash-Base model is positioned as a lightweight variant in StepFun's Step 3.5 series. The "Flash" designation indicates optimization for speed and reduced computational requirements compared to full-scale variants.
The model supports standard transformer architecture with SafeTensors format for optimized loading and inference. It is available for deployment across multiple regions, including US-based infrastructure.
Technical Specifications
The model is available on Hugging Face with the following characteristics:
- Format: SafeTensors (optimized tensor serialization)
- License: Apache 2.0 (permissive open-source)
- Architecture: Transformer-based text generation
- Pipeline: Text generation
As of release, the model has accumulated 58 likes and 135 downloads on Hugging Face, indicating early adoption from the community.
Research Background
The release includes references to two research papers (arxiv:2602.10604 and arxiv:2601.05593), suggesting the model incorporates recent algorithmic improvements from StepFun's research efforts.
Access and Availability
Step-3.5-Flash-Base is available for immediate download from Hugging Face. The Apache 2.0 license permits commercial use, modification, and distribution with appropriate attribution.
The model includes custom code implementations, indicating optimized inference kernels or specialized processing logic beyond standard transformer implementations.
What This Means
StepFun's release of Step-3.5-Flash-Base represents continued activity in the efficiency-focused segment of LLM development. The "Flash" branding suggests a deliberate positioning toward cost-effective inference—a key consideration for production deployments where computational overhead directly impacts operational costs. The open-source Apache 2.0 release indicates StepFun's strategy to build adoption through community distribution rather than API-gated access. Early download metrics suggest interest from practitioners seeking efficient alternatives to larger models.
Related Articles
Baidu open-sources ERNIE 5.5 Flash Base, an MoE model with 1M-token context for fine-tuning and inference
Baidu open-sourced ERNIE 5.5 Flash Base on November 5. It is a Mixture-of-Experts base model pretrained on 33.5 trillion tokens, with 128K pretraining context and up to 1M tokens in fine-tuning and inference. Baidu claims it sets state-of-the-art results among models of the same size.
Nous Research releases Arete-27B, a Qwen3.8-27B fine-tune that cuts propaganda rate from 22% to 0.9%
Nous Research fine-tuned Alibaba's Qwen3.8-27B on about 53 million tokens to remove Chinese state propaganda from its answers. According to Nous, the resulting Arete-27B model drops the propaganda rate from 22 percent to 0.9 percent while keeping general capabilities intact. Weights, test set and training data are public under Apache 2.0.
PrunaAI releases P-H3-Image: 8 MP image generation and editing in 8 steps, built on MiniMax-H3
PrunaAI has released P-H3-Image, an image model based on MiniMax-H3 that renders 8-megapixel images in 8 steps with no classifier-free guidance. It handles text-to-image and editing with up to 8 input images, and bf16 inference needs about 101 GB of GPU memory.
OpenCSG Agentic-27B: 27.36B-Parameter Tool-Use Model Runs on One A800, Commercial Use Needs Permission
OpenCSG has released Agentic-27B, a 27.36B-parameter model for multi-step tool use, built on the dense Qwen3.8-27B. The company reports it runs on a single 80GB NVIDIA A800, but commercial use requires authorization under the OpenCSG license, and all benchmark results are OpenCSG's own.
Comments
Loading...