Search the entire AI Orbit.

Living open-source code-review benchmark based on real-world pull-request behavior; F1 balances precision and re...

Controlled code-review benchmark across 50 hard-to-find bugs; F1 balances precision and recall.

Agents' Last Exam — official/provider-reported evaluation result. Methodology conditions are preserved in the re...

Human-validated 500-instance SWE-bench subset used to evaluate coding agents and systems.

Publicly reported professional agents benchmark result.

LiveCodeBench — official/provider-reported evaluation result. Methodology conditions are preserved in the result...

LiveCodeBench v6 — official/provider-reported evaluation result. Methodology conditions are preserved in the res...

Codeforces Percentile — official/provider-reported evaluation result. Methodology conditions are preserved in th...

Publicly reported finance agents benchmark result.

Publicly reported legal agents benchmark result.

Terminal-Bench 2.0 leaderboard accuracy for terminal-based agent tasks. Results are listed with agent, model, da...

Figure's public controlled evaluation of Helix S1 logistics manipulation after scaling demonstration data; score...

Accuracy on Exa's open WebCode Contents track measuring extraction fidelity against golden markdown.

Real-world production-time reduction calculated from an official customer example: two weeks to five days.

Internal Research Debugging Evaluation — official/provider-reported evaluation result. Methodology conditions ar...

Publicly reported coding & terminal agents benchmark result.

Toolathlon — official/provider-reported evaluation result. Methodology conditions are preserved in the result no...

Tau2 Average — official/provider-reported evaluation result. Methodology conditions are preserved in the result...

Publicly reported scientific agents benchmark result.

SWE-bench Verified (OpenAI 477 Tasks) — official/provider-reported evaluation result. Methodology conditions are...

Aider Polyglot (Diff) — official/provider-reported evaluation result. Methodology conditions are preserved in th...

SWE-Bench Pro — official/provider-reported evaluation result. Methodology conditions are preserved in the result...

Terminal-Bench 2.1 — official/provider-reported evaluation result. Methodology conditions are preserved in the r...

SWE-bench Verified (Anthropic Full 500) — official/provider-reported evaluation result. Methodology conditions a...

SWE-bench Verified (Anthropic Full 500 High Compute) — official/provider-reported evaluation result. Methodology...

Terminal-bench — official/provider-reported evaluation result. Methodology conditions are preserved in the resul...

SWE-bench Verified (Anthropic 489 Internal Subset) — official/provider-reported evaluation result. Methodology c...

SWE-bench Verified (Anthropic 489 High Compute) — official/provider-reported evaluation result. Methodology cond...

Aider Polyglot — official/provider-reported evaluation result. Methodology conditions are preserved in the resul...