LoL2LLM pulls your match history from the Riot Games API and reshapes it into a JSON form that ChatGPT and Claude can interpret without losing context. We do not run any AI inference on the server. The design intentionally hands inference to your own ChatGPT Plus or Claude Pro account, keeping the cost and privacy decisions on your side. The exported JSON includes lane-opponent deltas and role expectations, so the AI can evaluate lane quality and teamfight contribution — not just KDA.
Original articles on getting the most out of LoL2LLM — how to sharpen your AI prompts, which hidden metrics correlate with rank, and role-by-role evaluation frameworks.
Real ranked match data pulled from Faker's public account and run through AI analysis. The actual CS/min, damage share, and vision numbers of a pro, what they teach beginners, and every match ID so you can verify it yourself.
Real Challenger match data from Worlds-winning ADC Gumayusi's public account, run through AI analysis. The actual CS/min, kill participation, damage share, and control-ward counts of a pro ADC, with match IDs you can verify.
Real Grandmaster match data from T1 support Keria's public account, pulled via Riot's official API and run through AI analysis. The actual vision score, control ward counts, and kill participation of a pro support, plus what to track when CS/min doesn't apply, with match IDs you can verify.
Real Grandmaster match data from Gen.G top laner Kiin's public account, pulled via Riot's official API and run through AI analysis. The actual CS/min, kill participation, and damage share of a pro top laner, and why KP runs so much lower in the 'island' lane, with match IDs you can verify.
Browse all guides →
Is LoL2LLM affiliated with Riot Games?
No. LoL2LLM is an independent community tool that operates under the Riot Games third-party developer guidelines. See the disclaimer in the footer for full details.
Why copy-and-paste instead of automatic analysis?
So you can run inference on your own AI account (ChatGPT Plus, Claude Pro, etc.). Server-side LLM calls would force the site to monetize aggressively or restrict usage. Copy-paste keeps inference cost and privacy choices on your side.
Which AI is best for LoL analysis?
ChatGPT (GPT-4o, o1) excels at structured bulleted feedback. Claude (Sonnet, Opus) excels at long-form narrative reads of complex teamfights. Pasting the same JSON to both and comparing is a legitimate workflow.
What happens to my Riot account data?
Your Riot ID and PUUID are used only to call the Riot API; we don't persist them server-side. Search results are cached in your browser. See our Privacy Policy for full detail.
Which regions and queues are supported?
All major regions: JP, KR, NA, EUW, EUNE, and more. Most queues including Ranked Solo/Flex, Normal, ARAM, and Arena work.
What languages does LoL2LLM support?
The UI is available in Japanese, English, Chinese, Korean, and Spanish. Long-form guides and glossary are written in Japanese and English; other locales fall back to the English version.
I don't know what KDA / CS / vision score means.
The Glossary page explains 40+ core LoL terms. If you're new to reading match stats, start with the "How to Read Match Stats" guide and the glossary.
How much should I trust the AI's advice?
Conclusions derived directly from the JSON (CS deltas, damage share, etc.) are reliable. Meta or patch-balance opinions can be stale because of the model's knowledge cutoff. See the "Why AI Coaches Sometimes Give Bad Advice" guide.
KDA, vision score, damage share, gank, wave management — 40+ core LoL terms in one place. Skim the glossary before prompting your AI; the right vocabulary makes a noticeable difference in the answer.
Open the glossary →
Select the stats you want feedback on and paste directly into ChatGPT or Claude. AI will analyze your gameplay.
Search by Riot ID
Select and copy match
Paste to AI for analysis
This is an example. The real values are replaced with your own match data after you search.
Learn what each field meansarrow_forward