SkillEvolBench (No-Skill) - Acquisition Success: leaderboard
Coding-agent tasks from SkillEvolBench (Lei et al., 2026): 180 verifier-scored tasks in six environments such as code repair, API orchestration and document transformation, each model running in its vendor agent harness (Claude Code, Codex CLI or Gemini CLI). This board takes the no-skill control and scores the 90 acquisition-role tasks, the canonical, enriched and variant versions of each procedure; success rate in percent, higher is better.
Metric: Acquisition success rate (%; LSR, share of the 90 canonical, enriched and variant tasks solved with no skills or memory; verifier-scored). Source: arxiv.org. Saturation forecast: Around July 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.3 Codex | 44.4 | #114 |
| 2 | GPT-5.4 | 43.3 | #121 |
| 3 | GPT-5.2 Codex | 43.3 | #185 |
| 4 | Claude Opus 4.5 | 42.2 | #152 |
| 5 | Claude Sonnet 4.5 | 41.1 | #244 |
| 6 | Gemini 3.1 Pro (Preview) | 40 | #91 |
| 7 | Gemini 3 Flash | 40 | #159 |
| 8 | Claude Opus 4.6 | 38.9 | #100 |
| 9 | Claude Sonnet 4.6 | 37.8 | #158 |
| 10 | Gemini 2.5 Pro | 30 | #262 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=skillevolbench-no-skill-acquisition-success · How It Works · Data refreshed daily, snapshot 2026-11-12.