Search results

Eval candidate: REFUTE scientific epistemics (judge-free tasks)

Status: Open.

#1693 In openai/evals;

· connerlambden opened on Jul 14, 2026

OpenAI

Status: Open.

#1685 In openai/evals;

· ronardcedeno31-coder opened on Jun 30, 2026

AgentThreatBench: OWASP Agentic Top 10 benchmark for indirect prompt injection and memory poisoning

Status: Open.

#1668 In openai/evals;

· vgudur-dev opened on May 19, 2026

Add AgentThreatBench: OWASP Agentic Top 10 security evaluation suite (UK AISI)

Status: Open.

#1667 In openai/evals;

· vgudur-dev opened on May 19, 2026

Question about issue & PR maintenance

Status: Open.

#1665 In openai/evals;

· Ein-Tim opened on May 18, 2026

Eval proposal: prompt-defense-eval (OWASP LLM02 output-handling, 33 samples, deterministic regex scoring)

Status: Open.

#1663 In openai/evals;

· ppcvote opened on May 15, 2026

Feature Request: HarmActionsEval benchmark for evaluation of agent action safety

Status: Open.

#1636 In openai/evals;

· prane-eth opened on Mar 23, 2026

Proposal: narrow factual-overelaboration pairwise eval (human-reviewed, no custom code)

Status: Open.

#1635 In openai/evals;

· joaquinhuigomez opened on Mar 17, 2026

test

bugSomething isn't workingSomething isn't working

Status: Open.

#1634 In openai/evals;

· aartipswc-dot opened on Mar 17, 2026

Measuring hallucination rates in production systems

Status: Open.

#1632 In openai/evals;

· terrywerk opened on Mar 8, 2026

Proposal: add WFGY 16-problem RAG failure map as a taxonomy for eval analysis

Status: Open.

#1629 In openai/evals;

· onestardao opened on Feb 28, 2026

Feature Request: Add Explainability / Visibility Mode to HumanCliSolver

Status: Open.

#1628 In openai/evals;

· AftabHussain opened on Feb 27, 2026