Blog
Technical guides, architecture comparisons, and product updates for teams building production AI workflows.
Latest post
DeepSeek V4.1 Flash is now available on GraphN
Run DeepSeek V4.1 Flash on GraphN with a 1M-token context, image input, tool calling, a thinking toggle, and an OpenAI-compatible API.
Read latest post
Browse posts
Building AI That Knows Your Data: Inside GraphN Knowledgebases
Learn how GraphN Knowledgebases use vector search, multimodal embeddings, and reranking to ground AI workflows in your documents, images, and videos.
Bring your own Hugging Face model to GraphN without managing GPUs
How GraphN uses content-addressed caching, NVMe, and compile-artifact reuse for custom models. Cold wakes improved 1.7×–4.1× over our previous storage path, and a fully warm 131K Qwen model woke in 2m30s.
Qwen3.8-27B vs Claude Opus 4.6 on real document work
We tested Qwen's launch claim with a production document-extraction workflow: 8 documents, two conditions, field-level scoring, and public list prices for both models.
Qwen3.8-27B is now available on GraphN
Run Qwen3.8-27B on GraphN with 262K context, image and video understanding, thinking controls, tool calling, and an OpenAI-compatible API.
Publishing GraphN workflows: drafts, resources, and production versions
Learn how GraphN separates editable workflows from published versions, how linked resources fit the authoring model, and what to review before a production run.
Designing reliable async workflows in GraphN
Learn when to use asynchronous workflow runs, how to monitor their public lifecycle, and how to design retries, cancellation, and side effects safely.
Three trust boundaries for production agents: MCP tools, sandboxed functions, and grounded RAG
Choose the right GraphN boundary for external actions, isolated Python, and knowledge retrieval without confusing tool access with authority.
GraphN vs LangGraph: compare the production boundary, not just the graph
Compare GraphN and LangGraph across orchestration, state, tool isolation, deployment, observability, models, and migration boundaries.
GraphN vs Microsoft Agent Framework: decide where the platform begins
A technical comparison of GraphN, Microsoft Agent Framework, and Foundry Agent Service, including AutoGen migration, state, tools, identity, and operations.
GraphN vs Google ADK and Agent Platform: SDK composition or integrated workflow plane?
Compare GraphN with Google ADK and Gemini Enterprise Agent Platform across sessions, memory, tools, deployment, IAM, observability, and model boundaries.
GraphN vs CrewAI: where should agent autonomy stop and workflow control begin?
Compare GraphN with CrewAI and AMP across agent orchestration, Flow persistence, memory, tools, deployment, tracing, security, and migration.
GraphN vs Dify: compare who owns the application stack
Compare GraphN and Dify across visual authoring, agents, workflow state, code sandboxes, knowledge, model providers, deployment, security, and operations.
GraphN vs Amazon Bedrock AgentCore and Strands Agents: integrated product or composable AWS services?
Compare GraphN with AWS AgentCore and Strands Agents across orchestration, session isolation, memory, gateways, identity, observability, models, and operations.
Kanza AI's Clinical Reasoning System Launches with GraphN on NVIDIA Infrastructure in the US
GraphN, Lightning AI's agentic platform, powers Kanza AI's Clinical Reasoning System, now live and ranked first against Claude, GPT, and Gemini.