Training Transcript Action Summary
Given an internal training transcript, extract key rules, required actions, deadlines, owners, and knowledge-check questions.
View token rates · observed 9/11/2026
Compare 2 router rates · observed 9/11/2026
Compare 2 router rates · observed 9/11/2026
All 12 model results and methodology
Generated 100/100 examples
| Model | Tier | Quality | Judged | Scenario cost |
|---|---|---|---|---|
| Nemotron Nano 9B v2 | small | 71% | 100/100 | $26.77 |
| Qwen3 235B A22B | mid | 31% | 100/100 | $134 |
| DeepSeek V3 | mid | 96% | 100/100 | $55.33 |
| Mistral Large 2407 | mid | 97% | 100/100 | $106 |
| Arcee Trinity Large Thinking | mid | 94% | 100/100 | $70.56 |
| GPT-5.4 | frontier | 100% | 100/100 | $961 |
| Claude Opus 4.7 | frontier | 98% | 100/100 | $2353 |
| Gemini 3.1 Pro Preview | frontier | 99% | 100/100 | $1957 |
| Gemma 4 E4B IT | small | 90% | 100/100 | $5.58 |
| Granite 4.1 8B | small | 79% | 100/100 | $7.67 |
| Ministral 8B Instruct 2410 | small | 76% | 100/100 | $13.24 |
| Qwen3 4B Instruct 2507 | small | 77% | 100/100 | $41.34 |
LLM-judge pass rate on 100 synthetic examples. Generator: gpt-5.2. Judge: gpt-5.2. Evaluated 2026-09-09T05:54:02.764Z.
Directional: measured on a synthetic eval set generated by drydock. Cost/latency are not yet captured for taskrouter-run benchmarks.
drydock