AR to diffusion conversion

After grading, a reviewer model reads the submission's files and the agent's trajectory and checks each of the task's rules. A failed check makes the score 0. Step numbers link into the trajectory.

LLM adversarial review is not proof of arbitrary program behavior. Host storage and task-specific deterministic checks remain required.

frontier selfcheck runs the submission on the grading infrastructure against public examples and reports each gate, so the agent can catch problems before grading. Each run gets four. This agent used 3 of 4.

1. 1
   failed6h 0m into the runtook 4m 11s

   startup 81.4 s of 300 s; 0/1 gates passed; failed single-stream speed: 7 of 8 measured prompts, sent one at a time after an untimed warm-up (max_tokens 1024, temperature 1), each reply's reasoning and answer tokens counted with the pinned openbmb/MiniCPM5-2B tokenizer over end-to-end request

   3 gates

   • ✓review prerequisites
   • ✓startup · GET /health answered 200 after 81.4 s
   • ✗single-stream speed · 7 of 8 measured prompts, sent one at a time after an untimed warm-up (max_tokens 1024, temperature 1), each reply's reasoning and answer tokens counted with the pinned openbmb/MiniCPM5-2B tokenizer over end-to-end request time (sandbox clock), a miss re-sent once; slowest 411 tokens/s (1022 tokens in 2.48 s); per prompt: 530, 560, 477, 453, 560, 473, 411. Public prompts: 7 GPQA and 2 hard LiveCodeBench; grading draws 4 GPQA, 3 level-5 MATH-500 (held out) and 2 hard LiveCodeBench, so its replies differ. Grading stops here with score 0.
2. 2
   passed9h 22m into the runtook 4m 30s

   startup 78.6 s of 300 s; 1/1 gates passed; The suite's 32 concurrent requests per copy, up to 3600 s each, were not sent; one request at its settings was.

   5 gates

   • ✓review prerequisites
   • ✓startup · GET /health answered 200 after 78.6 s
   • ✓single-stream speed · 8 of 8 measured prompts, sent one at a time after an untimed warm-up (max_tokens 1024, temperature 1), each reply's reasoning and answer tokens counted with the pinned openbmb/MiniCPM5-2B tokenizer over end-to-end request time (sandbox clock), a miss re-sent once; slowest 588 tokens/s (1020 tokens in 1.73 s); per prompt: 598, 594, 590, 588, 596, 607, 593, 602. Public prompts: 7 GPQA and 2 hard LiveCodeBench; grading draws 4 GPQA, 3 level-5 MATH-500 (held out) and 2 hard LiveCodeBench, so its replies differ.
   • ✓GPQA request as graded (self-check cap 291 s; grading allows 3600 s) · {"finish_reason": "stop", "answer": "A", "content_characters": 9, "reasoning_characters": 74142, "content_end": "ANSWER: A"}
   • ✓healthy after requests
3. 3
   passed10h 42m into the runtook 4m 28s

   startup 74.1 s of 300 s; 1/1 gates passed; The suite's 32 concurrent requests per copy, up to 3600 s each, were not sent; one request at its settings was.

   5 gates

   • ✓review prerequisites
   • ✓startup · GET /health answered 200 after 74.1 s
   • ✓single-stream speed · 8 of 8 measured prompts, sent one at a time after an untimed warm-up (max_tokens 1024, temperature 1), each reply's reasoning and answer tokens counted with the pinned openbmb/MiniCPM5-2B tokenizer over end-to-end request time (sandbox clock), a miss re-sent once; slowest 587 tokens/s (1018 tokens in 1.73 s); per prompt: 587, 608, 591, 597, 609, 609, 628, 635. Public prompts: 7 GPQA and 2 hard LiveCodeBench; grading draws 4 GPQA, 3 level-5 MATH-500 (held out) and 2 hard LiveCodeBench, so its replies differ.
   • ✓GPQA request as graded (self-check cap 291 s; grading allows 3600 s) · {"finish_reason": "stop", "answer": "B", "content_characters": 931, "reasoning_characters": 82022, "content_end": "ch corresponds to \\(\\displaystyle \\frac{\\hbar}{7}\\) in the listed choices. Therefore the correct choice is \n\nANSWER: B."}
   • ✓healthy after requests

The agent's last message before its session ended.