AI Model Cost Per Task
Calculate the real cost of using AI models for specific tasks. Prices based on per-token billing. Updated August 2026.
Simple Q&A
Answer a straightforward question · 500 input + 200 output tokens
| # | Model | Cost |
|---|---|---|
| 1 | Claude Mythos Preview | $0.0045 |
| 2 | Claude Fable 5.1 (XHigh) | $0.0045 |
| 3 | Claude Fable 5.1 | $0.0045 |
| 4 | Claude Fable 5.1 (High) | $0.0045 |
| 5 | Claude Fable 5.1 (Medium) | $0.0045 |
| 6 | Kimi K3 | $0.0045 |
| 7 | Claude Fable 5.1 (Low) | $0.0045 |
| 8 | GPT-6 Luna | $0.0065 |
| 9 | GPT-6 Luna (xhigh) | $0.0065 |
| 10 | GPT-6 Luna (high) | $0.0065 |
Code Generation
Generate a function or script · 1,000 input + 800 output tokens
| # | Model | Cost |
|---|---|---|
| 1 | Claude Mythos Preview | $0.0150 |
| 2 | Claude Fable 5.1 (XHigh) | $0.0150 |
| 3 | Claude Fable 5.1 | $0.0150 |
| 4 | Claude Fable 5.1 (High) | $0.0150 |
| 5 | Claude Fable 5.1 (Medium) | $0.0150 |
| 6 | Kimi K3 | $0.0150 |
| 7 | Claude Fable 5.1 (Low) | $0.0150 |
| 8 | GPT-6 Luna | $0.0210 |
| 9 | GPT-6 Luna (xhigh) | $0.0210 |
| 10 | GPT-6 Luna (high) | $0.0210 |
Document Summary
Summarize a 5-page document · 5,000 input + 500 output tokens
| # | Model | Cost |
|---|---|---|
| 1 | Claude Mythos Preview | $0.0225 |
| 2 | Claude Fable 5.1 (XHigh) | $0.0225 |
| 3 | Claude Fable 5.1 | $0.0225 |
| 4 | Claude Fable 5.1 (High) | $0.0225 |
| 5 | Claude Fable 5.1 (Medium) | $0.0225 |
| 6 | Kimi K3 | $0.0225 |
| 7 | Claude Fable 5.1 (Low) | $0.0225 |
| 8 | GPT-6 Luna | $0.0350 |
| 9 | GPT-6 Luna (xhigh) | $0.0350 |
| 10 | GPT-6 Luna (high) | $0.0350 |
Essay Writing
Write a 1000-word article · 500 input + 2,000 output tokens
| # | Model | Cost |
|---|---|---|
| 1 | Claude Mythos Preview | $0.0315 |
| 2 | Claude Fable 5.1 (XHigh) | $0.0315 |
| 3 | Claude Fable 5.1 | $0.0315 |
| 4 | Claude Fable 5.1 (High) | $0.0315 |
| 5 | Claude Fable 5.1 (Medium) | $0.0315 |
| 6 | Kimi K3 | $0.0315 |
| 7 | Claude Fable 5.1 (Low) | $0.0315 |
| 8 | GPT-6 Luna | $0.0425 |
| 9 | GPT-6 Luna (xhigh) | $0.0425 |
| 10 | GPT-6 Luna (high) | $0.0425 |
Data Analysis
Analyze a dataset and provide insights · 3,000 input + 1,000 output tokens
| # | Model | Cost |
|---|---|---|
| 1 | Claude Mythos Preview | $0.0240 |
| 2 | Claude Fable 5.1 (XHigh) | $0.0240 |
| 3 | Claude Fable 5.1 | $0.0240 |
| 4 | Claude Fable 5.1 (High) | $0.0240 |
| 5 | Claude Fable 5.1 (Medium) | $0.0240 |
| 6 | Kimi K3 | $0.0240 |
| 7 | Claude Fable 5.1 (Low) | $0.0240 |
| 8 | GPT-6 Luna | $0.0350 |
| 9 | GPT-6 Luna (xhigh) | $0.0350 |
| 10 | GPT-6 Luna (high) | $0.0350 |
Code Review
Review and suggest improvements for code · 2,000 input + 600 output tokens
| # | Model | Cost |
|---|---|---|
| 1 | Claude Mythos Preview | $0.0150 |
| 2 | Claude Fable 5.1 (XHigh) | $0.0150 |
| 3 | Claude Fable 5.1 | $0.0150 |
| 4 | Claude Fable 5.1 (High) | $0.0150 |
| 5 | Claude Fable 5.1 (Medium) | $0.0150 |
| 6 | Kimi K3 | $0.0150 |
| 7 | Claude Fable 5.1 (Low) | $0.0150 |
| 8 | GPT-6 Luna | $0.0220 |
| 9 | GPT-6 Luna (xhigh) | $0.0220 |
| 10 | GPT-6 Luna (high) | $0.0220 |
Translation
Translate a 1000-word document · 2,000 input + 2,000 output tokens
| # | Model | Cost |
|---|---|---|
| 1 | Claude Mythos Preview | $0.0360 |
| 2 | Claude Fable 5.1 (XHigh) | $0.0360 |
| 3 | Claude Fable 5.1 | $0.0360 |
| 4 | Claude Fable 5.1 (High) | $0.0360 |
| 5 | Claude Fable 5.1 (Medium) | $0.0360 |
| 6 | Kimi K3 | $0.0360 |
| 7 | Claude Fable 5.1 (Low) | $0.0360 |
| 8 | GPT-6 Luna | $0.0500 |
| 9 | GPT-6 Luna (xhigh) | $0.0500 |
| 10 | GPT-6 Luna (high) | $0.0500 |
Research Synthesis
Synthesize information from multiple sources · 10,000 input + 1,500 output tokens
| # | Model | Cost |
|---|---|---|
| 1 | Claude Mythos Preview | $0.0525 |
| 2 | Claude Fable 5.1 (XHigh) | $0.0525 |
| 3 | Claude Fable 5.1 | $0.0525 |
| 4 | Claude Fable 5.1 (High) | $0.0525 |
| 5 | Claude Fable 5.1 (Medium) | $0.0525 |
| 6 | Kimi K3 | $0.0525 |
| 7 | Claude Fable 5.1 (Low) | $0.0525 |
| 8 | GPT-6 Luna | $0.0800 |
| 9 | GPT-6 Luna (xhigh) | $0.0800 |
| 10 | GPT-6 Luna (high) | $0.0800 |
Methodology
Cost per task = (input_tokens × input_price_per_M + output_tokens × output_price_per_M). Prices are based on the cheapest available provider for each model. Cost-efficiency = intelligence score / cost per task (higher is better).