We’ve removed the most significantrestrictionsonOpenAI Five’s gameplay—namely, wards, Roshan, and mirror match of fixed heroes, and will soon benchmark our progress by playing 99.95th-percentile Dota players. The OpenAI Five Benchmark match will be held12:30pm Pacific Time on August 5in San Francisco. The human team will includeBlitz(opens in a new window),Cap(opens in a new window),Fogged(opens in a new window), andMerlini(opens in a new window), some of whom are former professionals. The games will be streamed on ourTwitch(opens in a new window)channel andcasted(opens in a new window)by popular castersPurge(opens in a new window)andODPixel(opens in a new window).
Last year ourpreliminary Dota systemdefeated the world’s top professionals at the 1v1 version of Dota; last month OpenAI Five started defeating amateur teams at the full game of Dota (with somerestrictions). This event will show whether we have any hope of reaching the level of top professionals byThe International(opens in a new window))at the end of August.
We’ll kick off the OpenAI Five Benchmark with some warmup games against audience members; please let us know on theinvite request(opens in a new window)form if you’d like to volunteer.
Because our training systemRapidis very general, we were able to teach OpenAI Five many complex skills since June simply by integrating new features and randomizations. Many people pointed out that wards and Roshan were particularly important to include—and now we’ve done so. We’ve also increased the hero pool to 18 heroes. Many commenters thought these improvements wouldtake(opens in a new window)another(opens in a new window)year(opens in a new window). We’ll see how well these new game mechanics work on August 5th!
The latest version of OpenAI Five taking Roshan.
We will play with the following restrictions (the crossed out restrictions are those lifted since the original OpenAI Five blog post), which correspond to the last bits of the game we haven’t integrated:
We’ve increased the reaction time of OpenAI Five from 80ms to 200ms. This reaction time is much closer to human level, though we haven’t seen evidence of changes in gameplay as OpenAI Five’s strength comes more from teamwork and coordination than reflexes.
The final date for the match is August 5, and will run as follows:
## Want to attend in person?
If you’d like to attend the match in person in San Francisco, pleaserequest an invite(opens in a new window)by Friday July 20 at 11:59p PT; invites will be sent by the end of Monday July 23. Our venue is limited to 300 attendees, so we’ll be selecting invitees based on their answers to the request form. If you’ve already requested an invite, you’ll receive an email linking to this form as well.
View all company articles
Procgen and MineRL Competitions Company Jun 20, 2020
OpenAI Robotics Symposium 2019 Company Jun 5, 2019
OpenAI Five Finals Company Mar 26, 2019
Our Research * Research Index * Research Overview * Research Residency * OpenAI for Science * Economic Research
Latest Advancements * GPT-5.3 Instant * GPT-5.3-Codex * GPT-5 * Codex
Safety * Safety Approach * Security & Privacy * Trust & Transparency
ChatGPT * Explore ChatGPT(opens in a new window) * Business * Enterprise * Education * Pricing(opens in a new window) * Download(opens in a new window)
Sora * Sora Overview * Features * Pricing * Sora log in(opens in a new window)
API Platform * Platform Overview * Pricing * API log in(opens in a new window) * Documentation(opens in a new window) * Developer Forum(opens in a new window)
For Business * Business Overview * Solutions * Contact Sales
Company * About Us * Our Charter * Foundation * Careers * Brand
Support * Help Center(opens in a new window)
More * News * Stories * Livestreams * Podcast * RSS
Terms & Policies * Terms of Use * Privacy Policy * Other Policies
(opens in a new window)(opens in a new window)(opens in a new window)(opens in a new window)(opens in a new window)(opens in a new window)(opens in a new window)
OpenAI © 2015–2026 Manage Cookies
English United States
