StarCraft II: machine Grandmaster against human Ranked ladder
DeepMind’s AlphaStar reached Grandmaster on the public StarCraft II ladder on 30 October 2019, ranking above 99.8 per cent of ranked players with all three races, under camera-view observation, a capped action rate and an enforced reaction delay. It is a percentile result against the ladder population, not a match win over the single strongest human.
The record in full
| Discipline | StarCraft II REAL-TIME STRATEGY · LADDER PLAY |
|---|---|
| What is measured | Ranked ladder standing against the human player population |
| Human benchmark | Ranked ladder — The Battle.net ranked player population, European servers MULTINATIONAL · 30 OCTOBER 2019 · BLIZZARD BATTLE.NET LADDER |
| Machine benchmark | Grandmaster — AlphaStar 30 OCTOBER 2019 · BLIZZARD BATTLE.NET LADDER |
| Attributed to | Google DeepMind UNITED KINGDOM · UNITED STATES |
| Difference | No single comparable number. The machine passed the best humans in a contest with an unambiguous result. |
| Crossover | 30 October 2019 — AlphaStar reaches Grandmaster on the public ladder AlphaStar ranked above 99.8% of ranked human players on the European servers, playing all three races. |
| Verification status | VERIFIED Verified |
| Comparability status | BROAD Broadly comparable |
| Overtaq classification | CONFIRMED A machine passed the human benchmark in a contest or measurement Overtaq judges both well evidenced and legitimately comparable. |
| Data last checked | 23 AUGUST 2026 |
Conditions and comparability
Machine conditions
Played anonymously on the public ladder under human-like constraints: camera-view only, capped actions per minute, and a reaction delay. Reached Grandmaster with all three races, above 99.8% of ranked players.
Why this is broadly comparable
The constraints are what make this credible: camera-view observation, capped action rate and an enforced reaction delay, on the public ladder rather than a curated exhibition. It is a percentile result, not a match win over the single best human — Overtaq records it as such.
Machine progression
2 recorded results. Overtaq keeps every published result rather than only the current best, so the frontier can be read as a history. This table is the full data behind the chart on the interactive record.
| Date | Result | System | Status | Note |
|---|---|---|---|---|
| 24 January 2019 | Beats TLO and MaNa 10–1 | AlphaStar DeepMind |
REPORTED MODIFIED | An exhibition series against two professionals. The agent had full-map observation rather than a camera view, and MaNa won the one game played under camera constraints. |
| 30 October 2019 | Grandmaster, top 0.2% of ladder CROSSING | AlphaStar Final DeepMind |
VERIFIED BROAD | The crossover, under camera-view and action-rate constraints on the public ladder. |
Evidence
2 sources. Every source is a direct link to the primary record where one exists; where Overtaq has not been able to confirm an exact permanent URL, the entry says so rather than linking to a search page.
- Nature 575, 350–354 — Grandmaster level in StarCraft II using multi-agent reinforcement learning
30 OCTOBER 2019 · DOI:10.1038/s41586-019-1724-z
The Grandmaster result, the 99.8% percentile, and the camera and action-rate constraints.
- Google DeepMind — AlphaStar: Mastering the real-time strategy game StarCraft II
24 JANUARY 2019
The January 2019 exhibition results against TLO and MaNa and the observation conditions.
What is still missing
- No formal match has been played between a leading StarCraft II agent and the top-ranked human professional. The crossover is a ladder percentile, not a title.
Social cards
Every record is rendered into four formats in both themes from the same data, so the verification and comparability status travels with the numbers wherever the card is posted.
Related records
ALL RECORDS OVERTAKE TIMELINE HOW THIS IS VERIFIED SUBMIT A CORRECTION