Pwn2Own Berlin 2026 Closes With $1.3 Million in Prizes and 47 Zero-Days as DEVCORE Claims Master of Pwn
DEVCORE took the top prize with $505,000 and 50.5 points after three days at OffensiveCon, where 47 unique zero-days in Windows, Exchange, VMware, and AI tools earned researchers $1,298,250.
Overview
The annual Pwn2Own Berlin hacking competition concluded on May 16, 2026, with researchers collecting $1,298,250 for 47 unique zero-day vulnerabilities over three days, according to the Trend Micro Zero Day Initiative. The contest, held at the OffensiveCon conference from May 14 to May 16, targeted enterprise technologies and artificial intelligence, BleepingComputer reports. The total eclipsed last year’s $1,078,750 paid out across 28 zero-days, per the ZDI announcement for the competition.
Master of Pwn: DEVCORE
Taipei-based security firm DEVCORE claimed the Master of Pwn title with 50.5 points and $505,000 in total rewards, per the ZDI final results post. STARLabs SG finished second with 25 points and $242,500, and Out Of Bounds placed third with 12.75 points and $95,750.
The single highest prize of the competition — $200,000 — went to Cheng-Da Tsai, also known as Orange Tsai, of the DEVCORE Research Team, according to BleepingComputer. On Day Two, Orange Tsai chained three bugs to gain remote code execution with SYSTEM privileges on Microsoft Exchange, earning 20 Master of Pwn points alongside the cash award, as detailed by the ZDI Day Two blog post. A day earlier, the same researcher had earned $175,000 for a Microsoft Edge sandbox escape that chained together four logic bugs, BleepingComputer notes.
On the final day, splitline of the DEVCORE Research Team chained two bugs to exploit Microsoft SharePoint, earning $100,000 and 10 Master of Pwn points, per ZDI.
Day-by-Day Results
Day One was the most productive session of the contest. Competitors collected $523,000 in cash awards for 24 unique zero-days, according to the ZDI Day One blog post. Windows 11 was successfully compromised three times: Angelboy and TwinkleStar03 of DEVCORE exploited an improper access control flaw for $30,000, while Marcin Wiązowski used a heap-based buffer overflow for $15,000, and Kentaro Kawane of GMO Cybersecurity chained two use-after-free bugs for another $15,000.
On the AI side, Day One saw multiple AI products fall. Compass Security’s team of five exploited OpenAI Codex via a CWE-150 bug for $40,000. STARLabs SG’s Billy, Pan Zhenpeng, and Weiming Shi chained five bugs — including SSRF and code injection — to compromise LM Studio for $40,000. Independent researcher k3vg3n used three chained bugs involving SSRF and code injection to pwn LiteLLM for $40,000, all per the ZDI Day One results.
Valentina Palmiotti, known as “chompie,” of IBM X-Force Offensive Research collected $70,000 for rooting Red Hat Linux for Workstations and exploiting an NVIDIA Container Toolkit zero-day, BleepingComputer reports.
Day Two added $385,750 and 15 unique zero-days, bringing event totals to $908,750 with 39 unique vulnerabilities, according to the ZDI Day Two summary. Beyond Orange Tsai’s Exchange win, Viettel Cyber Security’s Le Duc Anh Vu exploited Cursor for $30,000. Sina Kheirkhah of the Summoning Team successfully exploited OpenAI Codex in a second round for $20,000. OtterSec’s Nikolaos Mourousias, Caue Obici, and Bruno Halltari used code injection to compromise LM Studio for $20,000. Ben Koo of Team DDOS exploited a use-after-free vulnerability to escalate privileges on Red Hat Enterprise Linux for $10,000. Among failures, Palo Alto Networks researchers and Rapid7’s Stephen Fewer were unable to complete their Apple Safari and Microsoft SharePoint exploits within the time constraints.
Day Three rounded out the competition with eight more zero-days. The biggest Day Three prize went to STARLabs SG’s Nguyen Hoang Thach, who used a Memory Corruption bug to exploit VMware ESXi with the Cross-tenant Code Execution add-on, earning $200,000 and 20 Master of Pwn points, per ZDI.
AI Targets Draw Record Participation
The 2026 edition expanded AI hacking into four distinct competition categories — AI Databases, Coding Agents, Local Inference, and NVIDIA products — as outlined in the ZDI competition announcement. Targets that fell included OpenAI Codex, LM Studio, LiteLLM, Ollama, Cursor, Anthropic Claude Code, and NVIDIA Megatron Bridge, SecurityWeek reports.
Demand was high enough to exceed available competition slots. According to SecurityWeek, “several teams were unable to sign up for Pwn2Own because all time slots were already taken,” with some researchers instead disclosing their findings directly to vendors.
What Happens Next
Under the ZDI’s coordinated disclosure policy, all vendors whose products were successfully compromised have been notified of the vulnerabilities. Vendors receive a 90-day window to develop and ship patches before ZDI publishes full technical details. The 47 new zero-days across Windows, Exchange, VMware ESXi, Red Hat Linux, and multiple AI platforms will translate into a wave of vendor security advisories in the months ahead.