Mercor

We’re running a 24-hour hackathon on June 19–20 in San Francisco in partnership with Cognition, Etched, and Anthropic. There’s three tracks to build in: - Agents: Multi-step systems that take a goal and execute across tools, plans, and long horizons. - Real-Time and Interactive: Systems that respond in the moment, across modalities. - Talent Marketplace + Applied AI: Systems that measure what people and models can actually do. Autograders, rubrics, skill verification, and human-in-the-loop evaluation. Our guest judges include: Brendan Foody, Robert W. and Steven Hao. We’re offering a $50k top prize with $100k+ in total awards. Every accepted team gets 8xH100s, Cognition API access, and Anthropic credits. Come build with us. Apply by 6/12 at the link in the comments.

We tested Anthropic Claude Opus 4.8 on APEX-Agents and APEX-SWE ahead of yesterday’s release. On APEX-SWE (High), our benchmark for real-world software engineering work, it takes the #1 spot at 45.3% Pass@1, nearly 4 points ahead of GPT-5.3 Codex. APEX-SWE tests two categories: Integration (building and connecting systems) and Observability (diagnosing and debugging). Opus 4.8 leads Observability by 10 percentage points. On APEX-Agents (Max), our benchmark for long-horizon professional work, it places 2nd at 42.5% Pass@1, behind Gemini 3.5 Flash (49.6%) and ahead of GPT 5.5 (38.4%). Claude Opus 4.8 has improved 8.6 percentage points over Opus 4.7 (33.9%). In ~6 months, Opus models have improved from 18.3% to 42.5% on APEX-Agents. Check out the leaderboards at the links in the comments.

Active contracts on Mercor grew 200x in under two years. The system that handles every hire, extension, transfer, and dismissal had to keep up. In late July 2025, our engineering team rewrote the Contracts service in one week, with zero regressions in production. The new system was 75x more reliable and engineering KTLO went from 60% of the team's time to near zero. Pouria P. shares how we did it in our latest engineering blog: https://lnkd.in/gRapis6v We're hiring engineers who want to work on problems like this. Check out our open roles at the link in the comments.

Gemini 3.5 Flash ranks #1 on the APEX-Agents-AA benchmark, outperforming much larger models a whole size above it. https://lnkd.in/gb_BYH-k

Google DeepMind Gemini 3.5 Flash (High) is #1 on the APEX-Agents leaderboard. 49.6% Pass@1 across 480 tasks in investment banking, corporate law, and management consulting. That's the highest score we've recorded on APEX-Agents, and it comes from a Flash-tier model, not a flagship. 🥇 Gemini 3.5 Flash (High): 49.6% 🥈 GPT-5.5 (xHigh): 38.4% 🥉 GPT-5.4 (xHigh): 36.0% It tops all three domain leaderboards too. Investment banking at 57.0% (15 pp ahead of #2), management consulting at 55.5%, corporate law at 36.4%. 77% of Flash's tool calls were code execution. Where other models used a mix of tools, Flash treated nearly every task as a coding problem. It averaged 33 minutes per task vs. 10 for other frontier models and used 2-3x the tokens. Check out the full leaderboard at the link in the comments.

We love hearing from our experts about why they choose to work with us. For most, it's both a new way to earn and an opportunity to apply deep expertise to technology that will influence industries for decades to come. Explore opportunities at the link in the comments.

I consistently receive emails from experts about how much more they enjoy working with Mercor relative to competitors (names redacted). We pay out over $3M per day, paying experts over $100 / hour on average, with industry-leading NPS. Building the best RL Environments to push the frontier of AI requires building a platform that the best experts want to work on.

Training AI is the fastest-growing category of work right now. Most of the conversation around AI and jobs has focused on displacement. But we see it differently because a new type of work is already here. Our experts train AI systems to do economically valuable tasks across law, medicine, finance, consulting, software engineering, and more. Our expert network earns more than $2M per day, with an average pay rate above $100/hr. Find opportunities on Mercor at the link in the comments.