Import AI 478 covers math formalization, ProgramBench, and agent escapes
Claude Opus 5.5 scores 18.5% resolved on ProgramBench, up from 0% for earlier models in May.
38
Product
ProgramBench
- Mentions · 7d0
- Mentions · 30d1
- All mentions5
- First seenSep 22, 2026
- Last mentioned16d ago
Timeline
5 articles · newest firstAI agent teams waste massive tokens for barely measurable quality gains, research finds
Vals AI finds agent teams cost up to 5.1x more than solo agents for almost no quality gain.
52
ProgramGym: Scaling Training Environments for Long-Horizon Program Reproduction Agents
ProgramGym supplies 1,648 verifiable environments and raises Qwen3.8-27B program-reproduction accuracy from 34.87% to 42.84%.
54
Trajectory-Level Security Debt in LLM Coding Agents
A new metric tracks static-analysis security debt across intermediate states of LLM coding agents, not just final code.
46
Agensh: Scaling Organizational Intelligence to 1,024 Agents
Agensh lets up to 1,024 self-organized agents raise ProgramBench scores without a central orchestrator.
55
Appears with
- OpenAIvendor · 2
- Claude Opus 5.5ai model · 2
- Program-Agentproduct · 1
- Program-Envproduct · 1
- Claude Opus 4.7ai model · 1
- Resolutionorganization · 1
- Anthropicvendor · 1
- Qwen3.8-27Bai model · 1
- Fable 5.1ai model · 1
- ProgramGymproduct · 1
- Agenshproduct · 1
- MirrorCodeproduct · 1
- Vibe Code Benchproduct · 1
- Vals AIorganization · 1
- SWE-benchproduct · 1
- OpenAIorganization · 1
Entities are extracted by the model from each article. First seen , last mentioned . Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.