A cross-provider analysis of LLM spending patterns across 15 common agent workflow tasks. Data curated from public pricing and benchmarks, July 2026.
| Task Type | Optimal Model Tier | Cost per 1K Tasks | Waste Rate |
|---|---|---|---|
| Text classification | Small (8B-class) | $0.12 | 71% |
| Entity extraction | Small (8B-class) | $0.18 | 65% |
| Summarization (short) | Medium (70B-class) | $0.84 | 44% |
| Code generation | Large (frontier) | $4.20 | 18% |
| Multi-step reasoning | Large (frontier) | $6.50 | 12% |
| Creative writing | Large (frontier) | $3.80 | 22% |
| Translation | Medium (70B-class) | $0.45 | 38% |
| Embedding generation | Specialized | $0.03 | 52% |
| Tool-use/function-calling | Large (frontier) | $5.20 | 28% |
| Document Q&A (RAG) | Medium (70B-class) | $1.10 | 41% |
| JSON extraction | Small (8B-class) | $0.15 | 68% |
| Intent detection | Small (8B-class) | $0.10 | 74% |
| Sentiment analysis | Small (8B-class) | $0.08 | 77% |
| Long-context analysis (>32K) | Large (frontier) | $8.40 | 15% |
| Image description | Large (frontier) | $2.10 | 33% |
Data provenance: Cost figures are curated from public LLM provider pricing pages (as of July 2026) and benchmark accuracy data from open evaluation suites (MMLU, HumanEval, MT-Bench). Waste rate is calculated as the percentage of teams using a model tier above the optimal for each task, based on a convenience sample of publicly shared agent workflow configurations from GitHub repositories, blog posts, and developer surveys.
Limitations: This is not a randomized controlled study. The sample skews toward teams that publicly share their setups, who may be more sophisticated than average. Actual costs vary with prompt length, context size, and provider-specific pricing changes.
"Optimal model tier" is defined as the smallest/cheapest model class that achieves ≥95% of frontier-model accuracy on standard benchmarks for that task type.