Built-in tool - verified
OpenAI Code Interpreter (Responses API containers)
OpenAI Code Interpreter (Responses API containers) is a model-provider built-in tool from OpenAI, read from its own pages on October 18, 2026. Isolation: Fully sandboxed virtual machine. Status: GA (inferred). 4 price rows, each in OpenAI Code Interpreter (Responses API containers)'s own unit.
Vendor: OpenAI. Docs: developers.openai.com. Pricing: developers.openai.com. Compare every agent sandbox
- Price rows
- 4
- Vendor claims
- 0
- Not stated
- 7
- Verified quotes
- 10
Pricing
In the provider's own unit, verbatim. Nothing is converted; the directory compares a price only with others in the same unit.
| Price | Unit, as stated | Quote and source | Notes |
|---|---|---|---|
| $0.03 | per 20-minute session per container, 1 GBDirectory block: Per 20-minute session per container | “1 GB $0.03, 4 GB $0.12, 16 GB $0.48, 64 GB $1.92 per 20-minute session per container.” | - |
| $0.12 | per 20-minute session per container, 4 GBDirectory block: Per 20-minute session per container | “1 GB $0.03, 4 GB $0.12, 16 GB $0.48, 64 GB $1.92 per 20-minute session per container.” | - |
| $0.48 | per 20-minute session per container, 16 GBDirectory block: Per 20-minute session per container | “1 GB $0.03, 4 GB $0.12, 16 GB $0.48, 64 GB $1.92 per 20-minute session per container.” | - |
| $1.92 | per 20-minute session per container, 64 GBDirectory block: Per 20-minute session per container | “1 GB $0.03, 4 GB $0.12, 16 GB $0.48, 64 GB $1.92 per 20-minute session per container.” | - |
- Free tier or credits
- Not stated
Note: Not stated.
Limits, lifecycle and isolation
- Max session duration
- Not stated
Note: Not stated beyond the 20-minute idle expiry and per-20-minute billing.
- Idle timeout and auto-pause
Container expires after 20 minutes unused; data discarded
“A container expires if it is not used for 20 minutes.”
developers.openai.com - Persistence and snapshots
No restore after expiry
“You can't move a container from an expired state to an active one.”
developers.openai.com - Isolation
Fully sandboxed virtual machine
“A container is a fully sandboxed virtual machine that the model can run Python code in.”
developers.openai.com - CPU and memory limits
Memory tiers 1g (default), 4g, 16g, 64g
“You can choose from 1g (default), 4g, 16g, or 64g.”
developers.openai.com - GPU
- Not stated
Note: Not stated.
- Egress controls
- Not stated
Note: Not stated in checked pages.
- SDKs and languages
Python execution only (tool called by the model)
“The Code Interpreter tool allows models to write and run Python code in a sandboxed environment”
developers.openai.com - Regions
- Not stated
Note: Not stated.
- Open source
- Not stated
Note: Not applicable.
- Self-host or BYOC
- Not stated
Note: Not applicable.
- Status
GA (inferred from published built-in tool pricing)inferred
“Hosted Shell and Code Interpreter”
Status inferred from - developers.openai.com
FAQ
How much does OpenAI Code Interpreter (Responses API containers) cost?
$0.03 per 20-minute session per container, 1 GB; $0.12 per 20-minute session per container, 4 GB; $0.48 per 20-minute session per container, 16 GB; $1.92 per 20-minute session per container, 64 GB. Prices are in the provider's own units as of October 18, 2026.
What isolation does OpenAI Code Interpreter (Responses API containers) use?
Fully sandboxed virtual machine.
How long can a session run on OpenAI Code Interpreter (Responses API containers)?
Not stated. Not stated beyond the 20-minute idle expiry and per-20-minute billing. Idle behavior: Container expires after 20 minutes unused; data discarded.
Does OpenAI Code Interpreter (Responses API containers) restrict outbound network access?
Not stated. Not stated in checked pages.
Sources
Last verified from each provider's own docs, pricing pages and repositories; every quote was string-matched against the page that day. Not a benchmark: prices are in the provider's own unit and change without notice, so check the linked source before you commit.
Newsletter
Know when agent infrastructure changes
One email per week. New model releases, agent tooling and what they mean for what you build.