Introducing Prospera: a benchmark that tests AI agents on real federal tax returns, designed by our research team in collaboration with CPAs and industry-leading tax professionals.
A complete federal return requires dozens of source documents, hundreds of interdependent calculations, and no room for errors. We evaluated GPT-5.4, Claude Opus 4.6, and Gemini 3.1 Pro with no hints on which forms to file, scored against 20+ expert-authored criteria per return.
Here’s the Results (Pass@3 ):
- GPT-5.4: 28.0%
- Gemini 3.1 Pro: 18.0%
- Claude Opus 4.6: 16.0%
To put those numbers in context, the tasks in Prospera weren't obscure edge cases. Filing a federal tax return is something millions of Americans do every year, yet 44% of evaluation criteria failed across all models.
Full report here: https://www.micro1.ai/benchmark/prospera
the micro1 robotics lab:
real world data for intelligent models that co-exist in the physical world.
we’re in-the-wild across 75 countries in 6,000+ unique environments collecting data. diverse movements, objects, and settings.
the future of AI is as human as you can imagine. join us to start training robots today (link in comments).
Excited to introduce micro1 Cortex, a contextual evaluation, visibility, and improvement platform for enterprise AI agents.
Foundational models are trained for general intelligence, but enterprises need agents that perform reliably inside their unique context: workflows, policies, data environments, and edge cases.
Cortex brings trust to enterprise AI by leveraging domain experts and real-world scenarios for any use case to test, diagnose, and improve how agents behave in production.
The U.S. MUST win the AI race
At micro1, we only work with U.S. AI labs and allied partners. This isn’t just about better products, it’s about who controls the intelligence layer of the global economy, and whether frontier AI strengthens the free world or empowers adversarial states.
AI will be the most important technology of our lifetime. It will automate not just software, but coordination, production, logistics, judgment, and execution. As automation compounds, output per worker rises, costs fall, and reshoring becomes economically viable, not through policy, but because intelligent systems operated domestically outperform global labor arbitrage.
That upside requires deep integration. To automate real-world enterprise or government functions, AI systems need full context, read and write access to core databases. There is no credible path to high-impact automation without granting frontier models that level of access.
If the U.S. does not win the AI race, organizations face an unacceptable choice: grant that access to Chinese models controlled by an adversarial government, or rely on inferior intelligence to automate functions that still must be automated. Both outcomes represent a major national security risk.
AI models reflect the human judgment embedded in their training, especially expert post-training data. When foreign actors dominate data creation, their priorities become encoded directly into the intelligence layer.
How we win: focus on software, massively scale structured expert human judgment, stop distillation, and have the U.S. government lead on rigorous AI evaluation so agents can be safely deployed.
This is a race to control the intelligence layer of the global economy.
And that must be US. 🇺🇸
The last mile doesn’t exist in AI.
@ali_ansarinik and Aaron, CEO of @box , share how automation creates a continuous loop of new functions and, in turn, new work across the AI economy.
This is how you get hired in 2026: build a real prototype for the company you want, send it to the exec team, and let the work speak for itself.
#ai #hiring #work #micro1 #jcal
Today we're excited to introduce micro1 Intelligence, the world’s most advanced platform for training frontier AI models.
Achieving AGI is bottlenecked by one main thing: high-quality data.
Data based on real-world environments that capture human expert workflows, complex decision-making, and reward signals models need to learn.
With micro1 Intelligence, frontier labs can train on RL environments across every subject matter all in one place.
