Most Expensive AI Models
Ranked from provider-published pricing · Prices checked 12 August 2026
The top of the price sheet, ranked by blended cost per million tokens. These are typically the largest reasoning models, sold for work where a wrong answer costs more than the tokens do.
Set against the cheapest list, the spread across this catalogue is several orders of magnitude, which is why picking a model by price alone, in either direction, tends to be a mistake.
The ranking
top 20 of 271Claude Opus 4.5 from Anthropic leads this ranking at $33.00/M.
| # | Model | Provider | Blended Cost per 1M Tokens | Input /1M | Output /1M | Context |
|---|---|---|---|---|---|---|
| 1 | Claude Opus 4.5 | Anthropic | $33.00/M | $15.00 | $75.00 | 200,000 |
| 2 | o1 | OpenAI | $28.50/M | $15.00 | $60.00 | 200,000 |
| 3 | Claude Fable 5 | Anthropic | $22.00/M | $10.00 | $50.00 | 1,000,000 |
| 4 | o3 | OpenAI | $19.00/M | $10.00 | $40.00 | 200,000 |
| 5 | GPT-4 Turbo | OpenAI | $16.00/M | $10.00 | $30.00 | 128,000 |
| 6 | GPT-5.6 Sol | OpenAI | $12.50/M | $5.00 | $30.00 | 1,050,000 |
| 7 | GPT-5.5 | OpenAI | $12.50/M | $5.00 | $30.00 | 1,050,000 |
| 8 | Claude Opus 5 | Anthropic | $11.00/M | $5.00 | $25.00 | 1,000,000 |
| 9 | Claude Opus 4.8 | Anthropic | $11.00/M | $5.00 | $25.00 | 1,000,000 |
| 10 | Claude Opus 4.7 | Anthropic | $11.00/M | $5.00 | $25.00 | 1,000,000 |
| 11 | Claude Opus 4.6 | Anthropic | $11.00/M | $5.00 | $25.00 | 1,000,000 |
| 12 | Grok 3 Fast | xAI | $11.00/M | $5.00 | $25.00 | 131,000 |
| 13 | Llama 3.1 405B (Rep) | Replicate | $9.50/M | $9.50 | $9.50 | 128,000 |
| 14 | GLM-4-Plus | Zhipu AI | $7.00/M | $7.00 | $7.00 | 128,000 |
| 15 | GLM-4V | Zhipu AI | $7.00/M | $7.00 | $7.00 | 2,000 |
| 16 | ERNIE 4.0 8K | Baidu | $6.72/M | $4.20 | $12.60 | 8,000 |
| 17 | Claude Sonnet 4.6 | Anthropic | $6.60/M | $3.00 | $15.00 | 1,000,000 |
| 18 | Claude Sonnet 4.5 | Anthropic | $6.60/M | $3.00 | $15.00 | 200,000 |
| 19 | Claude Sonnet 3.7 | Anthropic | $6.60/M | $3.00 | $15.00 | 200,000 |
| 20 | Claude Sonnet 3.5 | Anthropic | $6.60/M | $3.00 | $15.00 | 200,000 |
How this list is built
- Ranked most expensive first by a blended rate weighting input and output at 70/30. Default reference rate. 70% weight on input, 30% on output.
- Excludes embed models.
- Includes models marked available or preview; retired and deprecated SKUs are excluded.
- Prices are the rates each provider publishes, not estimates. How the blended rate is calculated.
Frequently asked questions
Why are some AI models so expensive?
Price tracks the compute a model burns per token. Large parameter counts and, increasingly, reasoning models that generate long internal chains of thought before answering both raise the cost of serving a single request. Providers also price by positioning: the top tier is sold on capability, not on being competitive per token.
Related rankings
Price it for your workload
Rankings are recomputed on every deploy from the catalogue. Prices checked 12 August 2026.