PulseAugur
EN
LIVE 11:45:30

StepFun releases 198B MoE vision-language model for agents

StepFun has released Step 3.7 Flash, a 198 billion parameter Mixture-of-Experts vision-language model designed for coding agents and search workflows. This new model features native multimodal understanding, improved tool-use reliability, and selectable reasoning depths to balance speed and computation. Step 3.7 Flash demonstrates significant performance gains on coding benchmarks like SWE-Bench Pro and offers an "Advisor Mode" that approaches Claude Opus 4.6 performance at a fraction of the cost. AI

IMPACT Sets a new benchmark for multimodal agentic coding performance and cost-efficiency, potentially influencing future agent development.

RANK_REASON New model release from a frontier lab (StepFun) with detailed technical specifications and benchmark results.

Read on Pandaily →

AI-generated summary · Google Gemini · from 6 sources. How we write summaries →

StepFun releases 198B MoE vision-language model for agents

COVERAGE [6]

  1. MarkTechPost TIER_1 English(EN) · Asif Razzaq ·

    StepFun Releases Step 3.7 Flash: A 198B MoE Vision-Language Model for Coding Agents and Search Workflows

    <p>StepFun releases Step 3.7 Flash, a 198B MoE model with native vision, 256k context, and Advisor Mode.</p> <p>The post <a href="https://www.marktechpost.com/2026/05/29/stepfun-releases-step-3-7-flash-a-198b-moe-vision-language-model-for-coding-agents-and-search-workflows/">Step…

  2. Pandaily TIER_1 English(EN) · [email protected] (Pandaily) ·

    Stepfun Open-Sources Step 3.7 Flash LLM Optimized for Agent Era

    Stepfun open-sources Step 3.7 Flash, a 196B-parameter sparse MoE LLM optimized for agent workflows with 400 tokens/s speed and native tool-calling capabilities.

  3. Mastodon — fosstodon.org TIER_1 한국어(KO) · [email protected] ·

    Notable news from an AI/Agent building perspective from Jason (@JasonVsTheNoise). StepFun has released the open-source model Step 3.7 Flash, emphasizing its fast speed and free availability, and it can be used on NousPortal. It can be utilized with agent models.

    Jason (@JasonVsTheNoise) AI/에이전트 구축 관점에서 주목할 만한 소식입니다. StepFun이 오픈소스 모델 Step 3.7 Flash를 공개했고, 빠른 속도와 무료 제공을 강조하며 NousPortal에서 사용할 수 있다고 합니다. 에이전트 모델과 함께 활용 가능하다는 점이 개발자 실무에 의미가 있습니다. https:// x.com/JasonVsTheNoise/status/2 061773074505113695 # ai # llm # opensource # agents # mod…

  4. Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] ·

    Stepfun has open-sourced Step 3.7 Flash, a 196B-parameter sparse MoE language model optimised for agent workflows. The model achieves generation speeds of up to

    Stepfun has open-sourced Step 3.7 Flash, a 196B-parameter sparse MoE language model optimised for agent workflows. The model achieves generation speeds of up to 400 tokens per second and features native multimodal understanding and tool-calling capabilities. Compatible with major…

  5. Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] ·

    Chinese AI startup StepFun has released Step 3.7 Flash, a 198bn parameter mixture-of-experts model for coding agents. Claims 56% on SWE-Bench Pro and Advisor Mo

    Chinese AI startup StepFun has released Step 3.7 Flash, a 198bn parameter mixture-of-experts model for coding agents. Claims 56% on SWE-Bench Pro and Advisor Mode reaching 97% of Claude Opus 4.6 performance at one-ninth the cost. Supports native multimodal inputs and 256k context…

  6. Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] ·

    Stepfun has open-sourced Step 3.7 Flash, a 196B-parameter sparse MoE language model optimised for agent workflows. The model achieves generation speeds of up to

    Stepfun has open-sourced Step 3.7 Flash, a 196B-parameter sparse MoE language model optimised for agent workflows. The model achieves generation speeds of up to 400 tokens per second and features native multimodal understanding and tool-calling capabilities. Compatible with major…