Skip to content
AgentsIndex

Semantic Kernel

A framework that turns prompts and APIs into function calls.

Best for
Developers who need to connect AI models to existing code and workflows.
Look elsewhere if
  • you want to use it for real without paying
GitHub stars
28.6k

What we checked

Pages last read
Facts checked
4 facts

From 6 sources, read 7 November 2026

AI answers
Too few read

58 buyer-question answers read, November 2026

At a glance

Tags
Open source

Code

Last push
9 October 2026

For developers

Works with
GitHub, Logic Apps, Azure Container Apps Dynamic Sessions

Semantic Kernel in our research

All research

Study

AI Tool Pricing Index, November 2026

Pricing

Free only

29 of the 361 tools we checked are free with no paid plan.

Edition price, checked 14 November 2026.

Pricing pages of the AI tools, November 2026 editionas of n = 361 tools checked

What is Semantic Kernel?

Semantic Kernel is a developer framework for building AI agents that turn prompts and existing APIs into function calls. It sits between the model and your code, using plugins and connectors to orchestrate workflows, with support for C#, Python, and Java plus observability, security, and filters. It integrates with GitHub, Logic Apps, and Azure Container Apps Dynamic Sessions.

It has been superseded by Microsoft's own successor SDK

As of this listing, the top of the Semantic Kernel GitHub README carries a flagged notice: "Semantic Kernel is now Microsoft Agent Framework! Microsoft Agent Framework (MAF) is the enterprise-ready successor to Semantic Kernel." Agent Framework reached a stable 1.0 release. Microsoft's Agent Framework devblog states it will "support Semantic Kernel v1.x for the foreseeable future" with critical bug fixes and security patches, and will bring some existing Semantic Kernel features to general availability, but says "the majority of new features" will be built for Agent Framework instead. The same devblog post recommends new projects start with Agent Framework. This isn't a hostile fork or a rumor — it's stated by Microsoft on the project's own primary channels.

What it actually is, for anyone still evaluating it

Semantic Kernel is a model-agnostic orchestration SDK for building AI agents and multi-agent systems, shipping as native packages for Python (PyPI: semantic-kernel), .NET (NuGet: Microsoft.SemanticKernel) and Java. It connects to OpenAI, Azure OpenAI, Hugging Face, NVIDIA NIM and others, supports local model runtimes (Ollama, LM Studio, ONNX), and includes a plugin system that accepts native code functions, prompt templates, OpenAPI specs, or MCP tools. It's the SDK Microsoft itself used internally before building Agent Framework on top of it, so the underlying orchestration concepts (kernel, plugins, planners) carry into the successor rather than being thrown away.

What it does

Semantic Kernel turns prompts and existing APIs into function calls, so an agent can act on real systems instead of only producing text. It sits between the model and your code: a request becomes a call through a plugin or a connector, and the result goes back into the model loop.

It ships connectors out of the box, works in C#, Python and Java, and has enterprise parts for observability, security and filters. Microsoft says it is used across Microsoft and by other Fortune 500 companies, and that the 1.0+ releases in all three languages avoid breaking changes, so a model can be swapped without rewriting the code around it. The docs name GitHub, Logic Apps and Azure Container Apps Dynamic Sessions as ways to extend agents and workflows (the Semantic Kernel docs).

Who it suits

  • Application developers who want agents to call the functions and services they already have.
  • Platform teams who need AI parts that fit an existing codebase.
  • Enterprise architects who need AI systems they can observe and secure at scale.
  • C#, Python and Java teams who want one framework across the three.
  • Builders who want AI to start workflows and business processes.

Development activity is real, not abandoned

The GitHub repo (microsoft/semantic-kernel) shows merged PRs landing daily — e.g. a .NET version bump to 1.80.0 merged 18 August 2026. NuGet's Microsoft.SemanticKernel package is on 1.80.0 (released 18 August 2026) with 15.4M total downloads; PyPI's semantic-kernel package is on 1.44.1 (uploaded 6 August 2026). The repo has roughly 400 contributors. This activity is consistent with Microsoft's stated 'critical bugs and security patches' commitment — it is not a project that has gone quiet, even though most new feature work is going to Agent Framework instead.

Two critical CVEs, both patched

OSV/GitHub Security Advisories list two CRITICAL-severity vulnerabilities disclosed in February 2026: CVE-2026-25592 (GHSA-2ww3-72rp-wpp4), an arbitrary file write in the .NET SessionsPythonPlugin via AI agent function calling, fixed in Microsoft.SemanticKernel.Plugins.Core 1.71.0; and GHSA-xjw9-4gw8-4rqx, a remote code execution issue in the InMemoryVectorStore filter functionality, fixed in 1.39.4. Both predate the current 1.80.0 (.NET) / 1.44.1 (Python) releases, so current installs are past the fixed versions — but anyone running an older pinned version should check they're above these thresholds.

Cost model

Semantic Kernel itself is free and open source under the MIT license — there is no vendor pricing tier to disclose. All cost is indirect: whatever LLM provider (OpenAI, Azure OpenAI, etc.) and vector store you connect it to.

Semantic Kernel price and terms

Terms

Free plan
YesYesThe SDK is free and open source under MIT with no usage limits
The terms in full

AI model calls are paid to the connected provider (OpenAI, Azure OpenAI and others).

For AI agents

Developer docs

Where these prices come from

learn.microsoft.com/en-us/semantic-kernel/overview, checked

What we checked on Semantic Kernel

YesPartlyNo

What it costs to startPartly

Free to use

Anyone can install it from pip or NuGet and use it without a sales call or signup, as an open-source kit.

Semantic Kernel is a lightweight, open-source development kit

learn.microsoft.com/en-us/semantic-kern…Checked

How you build with itYes

Public API

It is driven entirely from your own stack as an SDK, with plugins loadable from native code, prompts and OpenAPI specifications.

Semantic Kernel combines prompts with existing APIs to perform actions.

learn.microsoft.com/en-us/semantic-kern…Checked

How you build with itYes

MCP server

Publishing turns the agent into a hosted MCP server exposing one question-answering tool over streamable HTTP, with a Fabric bearer token per request.

When you publish a Fabric data agent, it becomes an MCP server that exposes a single MCP tool.

learn.microsoft.com/en-us/fabric/data-s…Checked

How you build with itYes

Open source

Microsoft describes it as an open-source development kit, and the source is public in its GitHub repository.

Our open-source GitHub repository is available for perusal and suggestions

learn.microsoft.com/en-us/semantic-kern…Checked

Semantic Kernel alternatives

Questions about Semantic Kernel

How much does Semantic Kernel cost?
Semantic Kernel has a free plan. From learn.microsoft.com/en-us/semantic-kernel/overview, checked 18 November 2026.
Does Semantic Kernel have an API or an MCP server?
Semantic Kernel has an API. From learn.microsoft.com/en-us/semantic-kernel/overview, checked 7 November 2026. Semantic Kernel has an MCP server. From learn.microsoft.com/en-us/fabric/data-science/data-agent-mcp-server, checked 4 November 2026.
Is Semantic Kernel open source, and can you host it yourself?
Semantic Kernel is open source. From learn.microsoft.com/en-us/semantic-kernel/support, checked 7 November 2026.
What have we checked on Semantic Kernel?
4 facts, under these headings: What it costs to start, how you build with it.
How do we check Semantic Kernel?
We hold 4 facts on Semantic Kernel, quoted from 6 pages. Facts last checked 7 November 2026. Price last checked 18 November 2026. Last change found 14 November 2026.

How we check every listing

Repository

Licence
MIT
Stars
28,639
Last push
9 October 2026

Semantic Kernel price history

We have read the price once so far, on 18 November 2026: free.

PricePlanFirst seenLast seenReadingsSource
Free18 November 202618 November 20261Pricing page, read 18 November 2026

From our research on 18 November 2026. A date is when we read the page, not when the price changed.

Similar to Semantic Kernel

Is Semantic Kernel yours?

Free, and the page then carries the verified owner mark. You confirm it with an email at learn.microsoft.com. Claim this listing

Report an error if a fact is wrong.