Large language models (LLMs) are now used in music education, and music educators are testing them for planning, feedback, and classroom communication. Music teaching relies on language for rehearsal notes, assessment prompts, and reflection, so these tools can shape daily instruction. This systematic review examines how LLMs are used in music education research and practice from 2019 to 2025. We report classroom and research uses, reported outcomes, and reporting gaps that limit repeat studies and evaluation. Prior work such as Rohwer (2024), Holster (2024), and O’Leary (2025) provides early guidance, but tools change fast, so an updated work is needed. We followed PRISMA guidelines. We searched ERIC, Scopus, and music education sources. We screened titles, abstracts, and full texts. We included studies that stated an LLM was used for a music education task in learning, teaching, or research data work. We extracted data on model and version, task type, music education setting, music focus, outcomes, checks, and ethics. The designs and outcomes differed across studies. Most included studies were published after late 2022. Four application areas appear across the set. These are research workflow support, analysis of interviews and open responses, questionnaire processing, and teaching and learning interventions. Within classrooms, studies reported LLM supports teaching tasks that rely on language and quick feedback. Teachers use chatbots to draft lesson plans, warm ups, rehearsal notes, rubrics, and short messages to families, and they revise them for fit and accuracy. AI tools can generate drills for rhythm, sight reading, and ear training, and some systems adjust difficulty based on student responses. Listening tools can give basic checks on pitch, rhythm, and tempo during practice, which helps when a teacher cannot hear every student. Students can use AI tools for composing and reflection drafts, but teachers should review artistic decisions in final performances. Common risks include hallucinated content, weak citations, and uneven cultural coverage. Studies also note academic integrity, authorship, privacy, and consent concerns when prompts include student work. Many reports on LLMs applications tend to focus on the research fields. The current moment as a snapshot of a fast-moving story, so classroom practice can change before studies are published. The use of LLMs in music classes can be informal and local. Teachers often share works and routines in workshops, districts, and online groups, and these do not become peer-reviewed studies.