Codex Extraction
Evaluates a model's ability to extract structured codex entries (characters, locations, objects, lore) from prose passages and return them as well-formed XML.
Long: The Spire of Echoes (Dense)
ToolingReasoning
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 | ||
|---|---|---|---|---|
| Gemini 3 Flash (Preview) | 98% | $0.0033 | 5.4s | |
| Mistral Small Creative | 95% | $0.0007 | 5.8s | |
| Grok 4 Fast | 96% | $0.0014 | 9.6s | |
| Mistral Medium 3.1 | 97% | $0.0035 | 7.5s | |
| Gemini 2.5 Flash | 93% | $0.0033 | 3.6s | |
| Mistral Small 3.2 24B | 93% | $0.0007 | 5.7s | |
| Qwen 3.5 Plus (2026-02-15) | 97% | $0.0041 | 14.9s | |
| Gemini 2.5 Flash Lite | 92% | $0.0005 | 2.4s | |
| Ministral 3 8B | 92% | $0.0008 | 4.9s | |
| Ministral 8B | 91% | $0.0005 | 5.3s | |
| Grok 4.1 Fast | 95% | $0.0018 | 17.9s | |
| Gemini 2.5 Flash Lite (Reasoning) | 95% | $0.0025 | 15.5s | |
| Z.AI GLM 4.5 | 96% | $0.0056 | 30.8s | |
| DeepSeek-V2 Chat | 95% | $0.0022 | 20.5s | |
| Mistral Large 3 | 94% | $0.0036 | 11.9s | |
| Ministral 3 3B | 89% | $0.0005 | 2.7s | |
| GPT-4.1 Mini | 92% | $0.0018 | 10.0s | |
| Ministral 3B | 89% | $0.0002 | 2.9s | |
| Claude Haiku 4.5 | 96% | $0.0096 | 6.6s | |
| Ministral 3 14B | 91% | $0.0010 | 9.9s | |
Most Stable Models (Top 20)
Ranked by stability (median × consistency). Click a model name to view its detail page.
| Score | Consistency | Stability | ||
|---|---|---|---|---|
| Gemini 3 Pro (Preview) | 99% | 100% | 98% | |
| Claude Opus 4 | 98% | 100% | 98% | |
| Gemini 3 Flash (Preview, Reasoning) | 99% | 99% | 98% | |
| Claude Opus 4.5 | 98% | 100% | 98% | |
| Claude Opus 4.6 | 98% | 100% | 98% | |
| Z.AI GLM 5 | 98% | 99% | 97% | |
| Claude Sonnet 4.6 (Reasoning) | 97% | 99% | 97% | |
| Gemini 3 Flash (Preview) | 98% | 99% | 97% | |
| Grok 4 | 97% | 99% | 97% | |
| Claude Opus 4.6 (Reasoning) | 98% | 99% | 97% | |
| Mistral Medium 3.1 | 97% | 100% | 96% | |
| Qwen 3.5 Plus (2026-02-15) | 97% | 99% | 96% | |
| o4 Mini High | 97% | 98% | 96% | |
| Aion 2.0 | 97% | 98% | 96% | |
| Claude Sonnet 4.6 | 97% | 99% | 96% | |
| GPT-5 | 96% | 98% | 95% | |
| Gemini 2.5 Pro | 97% | 98% | 95% | |
| Claude Haiku 4.5 | 96% | 99% | 95% | |
| Z.AI GLM 4.6 | 96% | 99% | 95% | |
| Claude 3.7 Sonnet | 96% | 99% | 95% | |
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 3 Flash (Preview) | 98% | $0.0033 | 5.4s | 97% | |
| Mistral Medium 3.1 | 97% | $0.0035 | 7.5s | 96% | |
| Mistral Small Creative | 95% | $0.0007 | 5.8s | 94% | |
| Grok 4 Fast | 96% | $0.0014 | 9.6s | 94% | |
| Qwen 3.5 Plus (2026-02-15) | 97% | $0.0041 | 14.9s | 96% | |
| Mistral Small 3.2 24B | 93% | $0.0007 | 5.7s | 91% | |
| Claude Haiku 4.5 | 96% | $0.0096 | 6.6s | 95% | |
| Gemini 2.5 Flash Lite | 92% | $0.0005 | 2.4s | 90% | |
| Ministral 3 8B | 92% | $0.0008 | 4.9s | 90% | |
| Gemini 3 Flash (Preview, Reasoning) | 99% | $0.011 | 19.1s | 98% | |
| Ministral 8B | 91% | $0.0005 | 5.3s | 90% | |
| Mistral Large 3 | 94% | $0.0036 | 11.9s | 91% | |
| Grok 4.1 Fast | 95% | $0.0018 | 17.9s | 92% | |
| Gemini 2.5 Flash Lite (Reasoning) | 95% | $0.0025 | 15.5s | 91% | |
| DeepSeek-V2 Chat | 95% | $0.0022 | 20.5s | 93% | |
| Ministral 3 3B | 89% | $0.0005 | 2.7s | 88% | |
| Gemini 2.5 Flash (Reasoning) | 95% | $0.0099 | 14.2s | 94% | |
| Ministral 3 14B | 91% | $0.0010 | 9.9s | 89% | |
| Gemini 2.5 Flash | 93% | $0.0033 | 3.6s | 85% | |
| GPT-4.1 Mini | 92% | $0.0018 | 10.0s | 88% | |
| Median | Evaluator | Top 3 | Flop 3 |
|---|---|---|---|
| 81.1% | Accuracy | ||
| 99.3% | Precision | ||
| 97.9% | Recall | ||
| 99.8% | Structural validity |