Published task benchmark

Churn Save Agent

Diagnose cancellation intent and offer only policy-allowed remedies before completing or retaining the account.

Compare measured quality, workload cost, and latency across model tiers. Results are task-specific and directional—not a general model ranking.

Sales and revenue operationsAnalysisEvaluated 9/10/2026100 evaluation examplesFallback pricing checked September 11, 2026
Analysis

Churn Save Agent

Diagnose cancellation intent and offer only policy-allowed remedies before completing or retaining the account.

Sales and revenue operationsPer account
Not benchmarked
No result on this tier
Token profile pending
/100K
Run this tier to add it
Gemma 4 E4B IT
E4B · Off-the-shelf
483 in · 601 out · 11.19s p50
$7.36 /100K
Current price · TrustedRouter
View token rates · observed 9/11/2026
TrustedRouter · $0.0211 input · $0.1055 output /1M
63%
Cheapest
Mistral Large 2407
123B · Off-the-shelf
523 in · 804 out · 22.20s p50
$155 /100K
Best current price · TrustedRouter
Compare 2 router rates · observed 9/11/2026
TrustedRouter · $0.5275 input · $1.58 output /1M
OpenRouter · $2.00 input · $6.00 output /1M
76%
GPT-5.4
Hosted · Off-the-shelf
422 in · 902 out · 9.05s p50
$1459 /100K
Best current price · OpenRouter
Compare 2 router rates · observed 9/11/2026
OpenRouter · $2.50 input · $15.00 output /1M
TrustedRouter · $2.64 input · $15.82 output /1M
100%
Best accuracy
All 12 model results and methodology

Generated 100/100 examples

ModelTierQualityJudgedScenario cost
Nemotron Nano 9B v2small34%100/100$30.31
Qwen3 235B A22Bmid34%100/100$195
DeepSeek V3mid46%100/100$72.75
Mistral Large 2407mid76%100/100$155
Arcee Trinity Large Thinkingmid40%100/100$100
GPT-5.4frontier100%100/100$1459
Claude Opus 4.7frontier94%100/100$4044
Gemini 3.1 Pro Previewfrontier77%100/100$1758
Gemma 4 E4B ITsmall63%100/100$7.36
Granite 4.1 8Bsmall33%100/100$8.81
Ministral 8B Instruct 2410small29%100/100$14.05
Qwen3 4B Instruct 2507small4%100/100$83.49

LLM-judge pass rate on 100 synthetic examples. Generator: gpt-5.2. Judge: gpt-5.2. Evaluated 2026-09-10T22:53:02.765Z.

Directional: measured on a synthetic eval set generated by drydock. Cost/latency are not yet captured for taskrouter-run benchmarks.

drydock

Have a neighboring workload?

Describe it in Task Explorer. If it is not measured yet, you can request a benchmark and follow future results.

Explore another task