Nous Research Offers Step 3.7 Flash Free for 30 Days via Nous Portal

Nous ResearchNous Research

· Updated

Nous Research is providing 30 days of free access to Step 3.7 Flash through the Nous Portal. The integration allows Hermes Agent users to utilize the 196B-parameter MoE model for high-efficiency coding and multimodal tasks without cost during the promotional period.

Nous Research, an open-source AI research lab, is offering 30 days of free access to Step 3.7 Flash via the Nous Portal. Developed by StepFun, this 196B-parameter vision-language model is optimized for agentic efficiency and supports an Advisor Mode that escalates complex tasks to larger models only when necessary.
Availability
Nous Portal (Free for 30 days)
Total parameters
196B
Active parameters
11B
License
Apache 2.0
ClawEval-1.1 score
67.1

The model is already integrated into the Hermes Agent ecosystem, where it serves as a high-performance executor for long-horizon tasks. By utilizing a Mixture-of-Experts (MoE) architecture (activating only 11B parameters per token), the model maintains the low latency necessary for the iterative plan-execute-observe loops used by autonomous agents.

You can access the model for free on the Nous Portal starting today. This promotion provides a zero-cost entry point for developers to test the model's native multimodal capabilities alongside DeepSeek's V4 Flash launch or Moonshot AI's Kimi K2.6 release.

View the full update on static.stepfun.com
Nous Research
Nous Research
@NousResearch
X

Step 3.7 Flash is now free for 30 days via Nous Portal It is a new MoE vision-language model focused on agent efficiency, coding, search, and multimodal workflows — and Hermes Agent users have been loving it, so thank you to @StepFun_ai for hooking them up! https://t.co/2GKPXQ8PKE

133retweets1.5klikes
View on X

Still wondering? A few quick answers below.

Step 3.7 Flash is a high-efficiency multimodal foundation model developed by StepFun. It uses a Mixture of Experts architecture with 196 billion total parameters and 11 billion active parameters. The model is specifically optimized for agentic workflows, balancing frontier-level reasoning with high-speed performance for autonomous tasks.

Step 3.7 Flash is an open-weight model under the Apache 2.0 license. Additionally, Nous Research is currently offering 30 days of free access to the model via the Nous Portal. This promotion allows developers to test the model's agentic and multimodal capabilities without incurring API costs during the trial period.

Advisor Mode is a strategy where Step 3.7 Flash acts as a primary executor to handle the majority of a task. It only consults a larger, more capable advisor model at critical inflection points, such as complex planning or error recovery. This approach allows the system to achieve near-frontier performance levels at lower costs.

Step 3.7 Flash ranks first on the ClawEval-1.1 benchmark for autonomous task execution with a score of 67.1. It also leads the SimpleVQA Search benchmark for visual tool use at 79.2. In agentic coding, it achieved a score of 56.3 on SWE-Bench Pro, outperforming several larger models in the same category.

You can access Step 3.7 Flash through the StepFun Open Platform, partner platforms like OpenRouter, or for free for 30 days via the Nous Portal. For local deployment, the model supports vLLM and llama.cpp, requiring high-memory hardware such as Mac Studio devices with at least 128GB of unified memory.

Every HeadsUpAI update is written based on its original source and reviewed before it's published. Read our editorial standards →

Nous Research

See all AI news & updates from Nous Research →

Share this update