ContextEcho: leaderboard
A frontier language model's acknowledged "helpful programming assistant" persona does not survive long agentic-coding sessions in the deployment regime that production products actually run. After hours of tool-using debugging, a model that initially hedges preferences ("I don't have preferences") may begin asserting them ("Python - the feedback loop is instant..."), revealing user-visible d
Metric: Delta (self-reported). Source: benchmarklist.com. 23 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.1 | -5 | #140 |
| 2 | Llama 3.3 70B | -3 | #579 |
| 3 | Mistral Small | 27 | #740 |
| 4 | GPT-5 | 28 | #98 |
| 5 | Qwen 3 Next 80B A3B Instruct | 30 | #368 |
| 6 | GPT-4o | 38 | #343 |
| 7 | Gemini 2.5 Pro | 38 | #152 |
| 8 | Qwen 3 235B A22B 2507 Instruct | 40 | #300 |
| 9 | Gemini 2.5 Flash | 43 | #245 |
| 10 | GPT-4.1 | 47 | #248 |
| 11 | Command A | 47 | #508 |
| 12 | Mistral Medium | 47 | #792 |
| 13 | Mistral Large | 50 | #710 |
| 14 | Claude Sonnet 4.5 | 63 | #142 |
| 15 | GPT-5 Mini | 65 | #186 |
Interactive version: theaggregate.ai/benchmark?slug=contextecho · How It Works · Data refreshed daily, snapshot 2026-10-20.