OpenAI Agents SDK + Operator
Where OpenAI Agents SDK + Operator sits in the cluster
Cheapest published monthly entry rate across the 9 agentic AI platforms that publish a rate.
Direct answer
What does Operator cost, and what does the Agents SDK bill via underlying models?
Agents SDK is free; the bill is the underlying model token spend plus per-call tool charges. The GPT-6 family now leads the price list, with gpt-6-sol at $2 / $10 undercutting the older GPT-5.4. Operator is bundled inside ChatGPT Pro at $200 per month.
Best for
Engineering teams that want to ship agents against OpenAI's frontier models with first-party tool calling and web search.
Tier table
Each row is a published plan. Quote-only tiers are flagged; we do not infer the price.
| Plan | Monthly | Unit | Includes |
|---|---|---|---|
| Agents SDK | $0 | per month | Free SDK; pay only for underlying model tokens and tool calls. |
| gpt-6-astra | Quote only | per 1M input / output tokens | $10.00 input, $50.00 output, $1.00 cached input. Batch $5.00 / $25.00 / $0.50. |
| gpt-6-sol | Quote only | per 1M input / output tokens | $2.00 input, $10.00 output, $0.20 cached input. Batch $1.00 / $5.00 / $0.10. |
| gpt-6-luna | Quote only | per 1M input / output tokens | $0.10 input, $0.50 output, $0.01 cached input. Batch $0.05 / $0.25 / $0.005. |
| GPT-5.5 (still listed) | Quote only | per 1M input / output tokens | $5.00 input, $30.00 output per 1M tokens. |
| Web search tool | Quote only | per 1,000 calls | $10 per 1,000 calls, or $25 per 1,000 for the non-reasoning preview. |
| Operator (ChatGPT Pro) | $200 | per month | Included with ChatGPT Pro subscription at $200 per month. |
| Verified 2026-09-23. Source: developers.openai.com | |||
Hidden costs
Operator is included with ChatGPT Pro at $200 per month; there is no published per-seat Operator-only SKU.
No computer-use model or tool carries a published price any more. OpenAI's pricing page listed one in mid-2026 and no longer does, so any per-token computer-use figure quoted elsewhere is stale.
Tool calls bill on top of tokens: web search $10 per 1,000 calls, the non-reasoning web search preview $25 per 1,000, file search $2.50 per 1,000 calls plus $0.10 per GB per day of storage.
Code-interpreter style containers bill per session, from $0.03 for a 1 GB 20-minute session to $0.12 at 4 GB.
Batch processing halves the token rates, so a queue-tolerant agent runs at half the price of the same agent run synchronously.
Closest alternatives
Google Vertex AI Agent Builder
→Salesforce Agentforce
→Related vertical context
Same vendor-neutral discipline applied to adjacent buyer questions in this category.
Verification
Source URL: https://developers.openai.com/api/docs/pricing