Claude Fable 5.1 vs GPT-5.4 Nano: cost per task
GPT-5.4 Nano is cheaper on 14 of 14 tasks and about 69.1x cheaper on average (geometric mean). Claude Fable 5.1 has the higher LMArena Text score (1502 vs 1401).
| Task | Claude Fable 5.1 | GPT-5.4 Nano | Difference |
|---|---|---|---|
| Answer a chat question | 6.3¢ | 0.15¢ | 42.8x |
| Reply to a customer email | 2.0¢ | 0.02¢ | 117.5x |
| Write a 1,500-word blog post | 18¢ | 0.30¢ | 61.8x |
| Write 10 product descriptions | 22¢ | 0.13¢ | 165.6x |
| Translate 1,000 words | 13¢ | 0.20¢ | 63.5x |
| Proofread one page (500 words) | 5.6¢ | 0.11¢ | 51.1x |
| Summarize a 20-page PDF | 22¢ | 0.33¢ | 68.3x |
| Extract data from an invoice | 5.2¢ | 0.08¢ | 66.3x |
| Analyze a 200-row spreadsheet | 85¢ | 0.44¢ | 192.1x |
| Answer from 5 retrieved documents (RAG) | 14¢ | 0.12¢ | 117.7x |
| Classify 100 support tickets | 4.5¢ | 0.15¢ | 28.8x |
| Review 500 lines of code | 40¢ | 0.75¢ | 53.2x |
| Write a SQL query | 5.5¢ | 0.10¢ | 54.5x |
| Run a 50-step coding agent | $3.46 | 8.7¢ | 39.8x |
Specs and list prices
| Claude Fable 5.1 | GPT-5.4 Nano | |
|---|---|---|
| Input per 1M | $10.00 | $0.20 |
| Output per 1M | $50.00 | $1.25 |
| Cached input per 1M | $0.25 | $0.02 |
| Context | 1M | 400K |
| Providers | 4 | 4 |
| Reasoning | always | default |
| LMArena Text | 1502 | 1401 |
| Released | Sep 1, 2026 | Mar 17, 2026 |
FAQ
Is Claude Fable 5.1 or GPT-5.4 Nano cheaper?
GPT-5.4 Nano is cheaper on 14 of 14 tasks and about 69.1x cheaper on average (geometric mean). Claude Fable 5.1 has the higher LMArena Text score (1502 vs 1401).
How much does Claude Fable 5.1 cost compared to GPT-5.4 Nano per token?
Claude Fable 5.1: $10.00 input and $50.00 output per 1M tokens. GPT-5.4 Nano: $0.20 input and $1.25 output. Token prices alone mislead - models use different numbers of tokens for the same task, which is what the task table shows.