Model reference · open weights

Step-3.5-Flash-Midtrain

Available as managed deployment LLMs stepfun-ai Text gen 1 variants 152 dl/mo

Step-3.5-Flash-Midtrain is an open-weight language model from stepfun-ai. AxForge deploys and operates it for you on dedicated EU-owned hardware — with the licence handled where one is required.

Available as managed deployment — configured and operated for you on dedicated EU hardware, quoted per deployment.

What it is

Makerstepfun-ai
TypeLanguage models
TaskText gen
Parameters (lead)197.8B
Context128k tokens
Runs withtransformers
Released2026-03-02
Popularity152 downloads / month
LicenceOpen weights

About

What Step-3.5-Flash-Midtrain is

1. Introduction

Step 3.5 Flash (visit website) is our most capable open-source foundation model, engineered to deliver frontier reasoning and agentic capabilities with exceptional efficiency. We also open-sourced the training codebase (SteptronOss), with support for continue pretrain, SFT, RL (WIP), and evaluation (WIP), and will open-source the SFT data. Built on a sparse Mixture of Experts (MoE) architecture, it selectively activates only 11B of its 196B parameters per token. This "intelligence density" allows it to rival the reasoning depth of top-tier proprietary models, while maintaining the agility required for real-time interaction.

2. Key Capabilities

  • Deep Reasoning at Speed: While chatbots are built for reading, agents must reason fast. Powered by 3-way Multi-Token Prediction (MTP-3), Step 3.5 Flash achieves a generation throughput of 100–300 tok/s in typical usage (peaking at 350 tok/s for single-stream coding tasks). This allows for complex, multi-step reasoning chains with immediate responsiveness.

  • A Robust Engine for Coding & Agents: Step 3.5 Flash is purpose-built for agentic tasks, integrating a scalable RL framework that drives consistent self-improvement. It achieves 74.4% on SWE-bench Verified and 51.0% on Terminal-Bench 2.0, proving its ability to handle sophisticated, long-horizon tasks with unwavering stability.

  • Efficient Long Context: The model supports a cost-efficient 256K context window by employing a 3:1 Sliding Window Attention (SWA) ratio—integrating three SWA layers for every full-attention layer. This hybrid approach ensures consistent performance across massive datasets or long codebases while significantly reducing the computational overhead typical of standard long-context models.

  • Accessible Local Deployment: Optimized for accessibility, Step 3.5 Flash brings elite-level intelligence to local environments. It runs securely on high-end consumer hardware (e.g., Mac Studio M4 Max, NVIDIA DGX Spark), ensuring data privacy without sacrificing performance.

3. Performance

Step 3.5 Flash delivers performance parity with leading closed-source systems while remaining open and efficient.

Performance of Step 3.5 Flash measured across Reasoning, Coding, and Agentic Abilities. Open-source models (left) are sorted by their total parameter count, while top-tier proprietary models are shown on the right. xbench-DeepSearch scores are sourced from official publications for consistency. The shadowed bars represent the enhanced performance of Step 3.5 Flash using Parallel Thinking.

Detailed Benchmarks

Benchmark# ShotsStep3.5 Flash (Base Midtrain)Step3.5 Flash (Base)MiMo‑V2 Flash (Base)GLM‑4.5 (Base)DeepSeek V3.1 (Base)DeepSeekV3.2 (Exp Base)Kimi‑K2 (Base)
# Activated Params-11B11B15B32B37B37B32B
# Total Params-196B196B309B355B671B671B1043B
General
BBH3-shot87.388.288.586.288.2†88.7†88.7
MMLU5-shot83.485.886.786.187.4†87.8†87.8
MMLU‑Redux5-shot87.289.290.6-90.0†90.4†90.2
MMLU‑Pro5-shot63.662.373.2-58.8†62.1†69.2
HellaSwag10-shot91.090.288.587.189.2†89.4†94.6
WinoGrande5-shot75.879.183.8-85.9†85.6†85.3
GPQA5-shot43.841.743.5*33.5*43.1*37.3*43.1*
SuperGPQA5-shot41.241.041.1-42.3†43.6†44.7
SimpleQA5-shot28.431.620.630.026.3†27.0†35.3
Mathematics
GSM8K8-shot88.988.292.387.691.4†91.1†92.1
MATH4-shot65.766.871.062.662.6†62.5†70.2
Code
HumanEval3-shot67.081.177.4*79.8*72.5*67.7*84.8*
MBPP3-shot79.079.481.0*81.6*74.6*75.6*89.0*
HumanEval+0-shot75.072.070.7-64.6†67.7†-
MBPP+0-shot62.470.671.4-72.2†69.8†-
MultiPL‑E HumanEval0-shot63.067.759.5-45.9†45.7†60.5
MultiPL‑E MBPP0-shot47.958.056.7-52.5†50.6†58.8
Chinese
C‑EVAL5-shot87.289.687.986.990.0†91.0†92.5
CMMLU5-shot86.988.987.4-88.8†88.9†90.9
C‑SimpleQA5-shot58.163.261.570.170.9†68.0†77.6
  1. “*” denotes cases where the original score was unavailable; we report results evaluated under the same test conditions as Step3.5 Flash for fair comparison.
  2. “†” indicates DeepSeek scores quoted from the MiMo‑V2‑Flash report.

Recommended Inference Parameters

  1. For general chat domain, we suggest: temperature=0.6, top_p=0.95
  2. For reasoning / agent scenario, we recommend: temperature=1.0, top_p=0.95.

4. Architecture Details

Step 3.5 Flash is built on a Sparse Mixture-of-Experts (MoE) transformer architecture, optimized for high throughput and low VRAM usage during inference.

4.1 Technical Specifications

ComponentSpecification
Backbone45-layer Transformer (4,096 hidden dim)
**Cont

From the published model card. Full card on the HuggingFace links in the sidebar.

Using it via the API

Call it like any OpenAI endpoint

Once AxForge deploys step-3-5-flash-midtrain for you, it answers on the OpenAI-compatible API — the same base URL and keys as every other model. (step-3-5-flash-midtrain below is illustrative; you get the exact model name on deployment.)

$ curl -sS https://api.axforge.ai/v1/chat/completions \
  -H "Authorization: Bearer $AXFORGE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"step-3-5-flash-midtrain","messages":[{"role":"user","content":"Hello"}]}'

Create an account — your API key is available in the console. 5M tokens/month currently included with every new account at launch.

© 2026 AxForge · EU-hosted AI infrastructure Pricing Docs Trust Privacy Terms