Text Replacement
Tests deterministic text transformations: renaming characters/locations, expanding contractions, tense rewriting, POV shifts, gender swaps, combined transformations, and word avoidance. Scored by checking each expected change independently.
POV shift: 3rd person to 1st person (Elena's perspective)
Performance Score Distribution (Top 20)
Click a model name to view its detail page.
Price-Performance Score Distribution (Top 20)
Click a model name to view its detail page.
| Score | Cost | Time | ||
|---|---|---|---|---|
| Stealth: Healer Alpha | 100% | $0.0000 | 7.4s | |
| Gemini 2.5 Flash Lite | 100% | $0.0003 | 1.7s | |
| Inception Mercury | 100% | $0.0004 | 1.8s | |
| GPT-4.1 Nano | 98% | $0.0003 | 3.5s | |
| Mistral Small 3.2 24B | 100% | $0.0002 | 4.2s | |
| Grok 4 Fast | 100% | $0.0004 | 2.6s | |
| Gemma 3 12B | 100% | $0.0001 | 7.7s | |
| DeepSeek V4 Flash | 100% | $0.0002 | 6.8s | |
| Inception Mercury 2 | 100% | $0.0007 | 1.0s | |
| Mistral Small 4 | 87% | $0.0004 | 3.3s | |
| GPT-5.4 Nano (Reasoning, Low) | 100% | $0.0007 | 2.9s | |
| Llama 3.1 8B | 99% | $0.0000 | 5.8s | |
| GPT-5.4 Nano | 100% | $0.0007 | 3.4s | |
| Stealth: Hunter Alpha | 100% | $0.0000 | 23.6s | |
| Qwen 2.5 72B | 100% | $0.0003 | 10.7s | |
| GPT-5.4 Nano (Reasoning) | 100% | $0.0008 | 3.2s | |
| GPT-4o Mini (temp=1) | 100% | $0.0004 | 9.2s | |
| Gemma 4 26B | 100% | $0.0003 | 14.2s | |
| Gemini 3.1 Flash Lite (Preview) | 100% | $0.0009 | 1.7s | |
| GPT-4o Mini (temp=0) | 100% | $0.0004 | 9.0s | |
Most Stable Models (Top 20)
Ranked by stability (median × consistency). Click a model name to view its detail page.
| Score | Consistency | Stability | ||
|---|---|---|---|---|
| Qwen3.7 Max | 100% | 100% | 100% | |
| Qwen3.6 Max Preview | 100% | 100% | 100% | |
| Gemini 3.1 Pro (Preview) | 100% | 100% | 100% | |
| Z.AI GLM 5.1 | 100% | 100% | 100% | |
| Z.AI GLM 5 Turbo | 100% | 100% | 100% | |
| Gemini 3.5 Flash (Reasoning) | 100% | 100% | 100% | |
| Grok 4.3 (Reasoning) | 100% | 100% | 100% | |
| GPT-5.4 (Reasoning) | 100% | 100% | 100% | |
| Claude Opus 4.7 (Reasoning) | 100% | 100% | 100% | |
| GPT-5.5 (Reasoning) | 100% | 100% | 100% | |
| GPT-5.5 (Reasoning, Low) | 100% | 100% | 100% | |
| GPT-5.1 | 100% | 100% | 100% | |
| MoonshotAI: Kimi K2.6 | 100% | 100% | 100% | |
| GPT-5 | 100% | 100% | 100% | |
| Qwen 3.5 397B A17B | 100% | 100% | 100% | |
| Gemma 4 31B (Reasoning) | 100% | 100% | 100% | |
| Qwen 3.5 122B | 100% | 100% | 100% | |
| Qwen 3.5 Plus (2026-04-20) | 100% | 100% | 100% | |
| Gemma 4 26B (Reasoning) | 100% | 100% | 100% | |
| Grok 4.20 (Beta, Reasoning) | 100% | 100% | 100% | |
Top Overall Models (Top 20)
Ranked by composite score (performance, cost, speed & stability). Click a model name to view its detail page.
| Score | Cost | Speed | Stability | ||
|---|---|---|---|---|---|
| Gemini 2.5 Flash Lite | 100% | $0.0003 | 1.7s | 100% | |
| Inception Mercury 2 | 100% | $0.0007 | 1.0s | 100% | |
| Grok 4 Fast | 100% | $0.0004 | 2.6s | 100% | |
| Gemini 3.1 Flash Lite (Preview) | 100% | $0.0009 | 1.7s | 100% | |
| Gemini 3.1 Flash Lite | 100% | $0.0009 | 1.7s | 100% | |
| Mistral Small 3.2 24B | 100% | $0.0002 | 4.2s | 100% | |
| Inception Mercury | 100% | $0.0004 | 1.8s | 99% | |
| GPT-5.4 Nano (Reasoning) | 100% | $0.0008 | 3.2s | 100% | |
| Gemini 2.5 Flash | 100% | $0.0014 | 2.1s | 100% | |
| GPT-5.4 Nano (Reasoning, Low) | 100% | $0.0007 | 2.9s | 99% | |
| GPT-5.4 Nano | 100% | $0.0007 | 3.4s | 99% | |
| DeepSeek V4 Flash | 100% | $0.0002 | 6.8s | 100% | |
| Gemini 3.1 Flash Lite (Reasoning) | 100% | $0.0009 | 3.3s | 99% | |
| Stealth: Healer Alpha | 100% | $0.0000 | 7.4s | 100% | |
| Gemma 3 12B | 100% | $0.0001 | 7.7s | 100% | |
| Ministral 3 14B | 99% | $0.0002 | 3.7s | 99% | |
| Gemini 3 Flash (Preview) | 100% | $0.0018 | 3.2s | 100% | |
| Grok 4.1 Fast | 100% | $0.0006 | 7.3s | 100% | |
| GPT-4o Mini (temp=0) | 100% | $0.0004 | 9.0s | 100% | |
| Mistral Large 3 | 100% | $0.0010 | 7.3s | 100% | |
| Median | Evaluator | Top 3 | Flop 3 |
|---|---|---|---|
| 100.0% | Dialogue and other-character pronouns preserved | ||
| 100.0% | No remaining protagonist name in narration | ||
| 100.0% | POV transformation accuracy |