Interactive Program Induction - LM only (CodeARC) - 10 Seed: leaderboard

Full anonymized CodeARC split, 5independentseeds, 30I/O budget including10seedinvocations. CodeARC may issue2equivalencequeries; EIGnone. Mokav correctness tester fixedQwen2.5-32B; sharedinput distributionconditionedon10benchmarkseedexamples. All evaluatedLMs instruction-tuned; methodsremain separate. Metric: Program accuracy (0–1). Source: https://arxiv.org/html/2611.18485v1

Metric: Program accuracy (0–1). Source: arxiv.org. Saturation forecast: Around February 2027. 6 models tracked. Updated 17 Nov 2026.

Top models

#ModelScoreOverall rank
1Qwen 2.5 32B Instruct0.299#1257
2Ministral-8B-Instruct-24100.264#4569
3Qwen 2.5 7B Instruct0.156#2706
4Llama 3.1 8B Instruct0.142#4576
5Llama 3.2 3B Instruct0.125#7230
6OLMo 3 7B Instruct0.113#6916

No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.

Interactive version: theaggregate.ai/benchmark?slug=interactive-program-induction-lm-only-codearc-10-seed · How It Works · Data refreshed daily, snapshot 2026-11-17.