Model prices
The prices on this page apply to legacy pay-as-you-go spaces. New spaces are on the subscription model.
Agent models​
Every Agent task uses credits based on its complexity and choice of Agent model. Small and short tasks cost little, while bigger and longer tasks cost more. The following table shows the average cost per 100 messages for every Agent model available in GPT for Sheets and GPT for Excel. The values are based on recent usage data.
| Provider | Model | Average cost per 100 messages |
|---|---|---|
 OpenAI | GPT 5.5 | $146 |
| OpenAI | GPT 5.4 | $45 |
| OpenAI | GPT 5.1 | $15 |
 Anthropic | Opus 4.8 | N/A |
| Anthropic | Opus 4.7 | $99 |
| Anthropic | Opus 4.6 | $99 |
Bulk models​
Bulk models used by GPT for Work are priced differently depending on whether you use an API key or not:
-
Without an API key: The price per million tokens covers everything. Models that support prompt caching will get a 75% discount on cached input tokens.
-
With an API key: You pay GPT for Work the price per million tokens and also pay the AI provider directly for API usage.
The following tables show the prices for all bulk models available in GPT for Work, without and with an API key.
Models without an API key​
The following table shows the price per million consumed tokens for every bulk model available without an API key. For GPT for Sheets and GPT for Excel, the table also shows the average cost per 1,000 processed rows (based on recent usage data).
| Provider | Model | Price | Avg. cost per 1K rows (spreadsheets) |
|---|---|---|---|
| OpenAI | gpt-5.5 | $50 / 1M tokens | N/A |
| OpenAI | gpt-5.4 | $50 / 1M tokens | $25.27 |
| OpenAI | gpt-5.4-mini | $9 / 1M tokens | $3.41 |
| OpenAI | gpt-5.4-nano | $2.50 / 1M tokens | $1.65 |
| OpenAI | gpt-5.2 | $50 / 1M tokens | $32.21 |
| OpenAI | gpt-5.1 | $40 / 1M tokens | $23.47 |
| OpenAI | gpt-5-mini | $8 / 1M tokens | $6.75 |
| OpenAI | gpt-5-nano | $1 / 1M tokens | $1.18 |
| OpenAI | gpt-4.1 | $30 / 1M tokens | $11.23 |
| OpenAI | gpt-4.1-mini | $6 / 1M tokens | $2.36 |
| OpenAI | gpt-4.1-nano | $1 / 1M tokens | $0.45 |
| OpenAI | gpt-5-search-api | $3 / 1M input tokens $13 / 1M output tokens + $15 / 1K searches | $62.03 |
| OpenAI | gpt-4o-search-preview | $3 / 1M input tokens $13 / 1M output tokens + $30 / 1K searches | $26.34 |
| OpenAI | gpt-4o-mini-search-preview | $2 / 1M tokens + $30 / 1K searches | $25.64 |
| OpenAI | gpt-4o | $30 / 1M tokens | $7.48 |
| OpenAI | gpt-4o-mini | $2 / 1M tokens | $0.98 |
| OpenAI | gpt-5 | $40 / 1M tokens | $27.41 |
| OpenAI | gpt-5-chat | $20 / 1M tokens | $7.03 |
| OpenAI | gpt-4-turbo | $35 / 1M tokens | $24.89 |
| OpenAI | o1 | $70 / 1M tokens | $26.81 |
| OpenAI | o3-mini | $10 / 1M tokens | N/A |
| OpenAI | o3 | $10 / 1M tokens | $6.48 |
| OpenAI | o4-mini | $10 / 1M tokens | $4.03 |
| OpenAI | gpt-3.5-turbo | $2.50 / 1M tokens | $1.46 |
 Perplexity | sonar | $10 / 1M tokens + $10, $13, or $17 / 1K searches | $8.88 |
| Perplexity | sonar-pro | $20 / 1M tokens + $13, $15, or $19 / 1K searches | $15.78 |
 Google | gemini-2.5-flash | $4 / 1M tokens + $40 / 1K searches | $15.56 |
| Anthropic | claude-4-6-sonnet | $30 / 1M tokens | $16.12 |
| Anthropic | claude-4-sonnet | $20 / 1M tokens | $25.39 |
| Anthropic | claude-4.5-sonnet | $20 / 1M tokens | $12.86 |
| Anthropic | claude-4.7-opus | $50 / 1M tokens | N/A |
| Anthropic | claude-4.6-opus | $50 / 1M tokens | $10.26 |
| Anthropic | claude-4.5-opus | $50 / 1M tokens | $179.35 |
| Anthropic | claude-4.5-haiku | $6 / 1M tokens | $3.99 |
 Parallel | speed | $10 / 1K searches | $7.00 |
Models with an API key​
The following table shows the price per million consumed tokens for bulk models with an API key. Models priced at the standard $1 per million tokens are grouped into a single summary row.
| Provider | Model | Price |
|---|---|---|
| OpenAI | gpt-4o-mini-search-preview | $0.30 / 1M tokens |
| OpenAI | gpt-4o-mini | $0.30 / 1M tokens |
| OpenAI | gpt-4o-mini (fine-tuned) | $0.30 / 1M tokens |
| Varies | All other models | $1 / 1M tokens |
-
To estimate your credit usage based on common use cases, try our estimation tools.
-
For examples of token consumption for different features in the add-ons, see How credits are consumed.