> ## Documentation Index
> Fetch the complete documentation index at: https://docs.braintrust.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Plans and limits

> Choose the right Braintrust plan for your needs

Braintrust offers three plans designed to meet the needs of different types of users, from individual developers to large enterprises. The **Starter** plan is free with generous usage limits, while **Pro** and **Enterprise** plans offer higher limits, advanced features, and enhanced security and compliance capabilities.

## Plans

For detailed pricing and to estimate your monthly costs, visit the [Pricing page](https://www.braintrust.dev/pricing).

* **Starter** (\$0 platform fee) — Start building and evaluating AI applications for free with generous usage limits. No credit card required.
  * Best for individual developers and small teams.
  * On-demand usage pricing applies when your usage grows beyond the free-tier.
* **Pro** (\$249/month) — Everything in Starter, plus advanced observability, security features, and higher usage limits.
  * Best for growing teams and production applications.
  * Lower on-demand usage rates than Starter.
* **Enterprise** (Custom pricing) — Everything in Pro, plus custom usage limits, advanced security and compliance features, flexible deployment options (BYOC and self-hosted), and dedicated support.
  * Best for large organizations with enterprise requirements.
  * Annual invoice.

> [!TIP]
>
> Qualifying early-stage startups can apply for 6–12 months of free Pro plan access. To be eligible, you must be a new Braintrust customer at Series A or earlier and have raised at least \$100K in funding. Visit the [startups page](https://www.braintrust.dev/startups) to apply.

### Usage limits

Usage limits determine how much data you can log, score, and retain, and how much built-in model usage is included with your plan. On-demand rates apply once you exceed the included limits.

| Limit | Starter | Pro | Enterprise |
| - | - | - | - |
| Processed data (The total bytes of data ingested across logs, experiments, and datasets. Includes inputs, outputs, prompts, metadata, traces and spans, datasets, attachments, and any other related information. Measured at ingestion, so deleting data doesn't reduce it.) | 1 GB / month (\$4/GB after) | 5 GB / month (\$3/GB after) | Custom |
| Scores (The number of scores recorded across offline and online evaluations and human reviews.) | 10k / month (\$2.50/1k after) | 50k / month (\$1.50/1k after) | Custom |
| Data retention (How long your logs and playground data are retained. Data outside your retention window isn't available in the UI or returned by SQL and API queries. Experiments have up to 365-day retention on all plans. Pro plans can extend retention up to 180-day retention for an additional storage charge. Enterprise retention is set with your account team, up to 365-day retention.) | 14-day retention | 30-day retention (\$0.50/GB/mo after) | Custom |
| Model credits (Monthly credits for built-in model usage, including Topics. Unused credits do not roll over to the next month. After monthly credits are exhausted, on-demand token rates apply.) | \$10 / month | \$100 / month | Custom |

> [!TIP]
>
> Pro plan on-demand usage rates are lower and provide better economics as your usage scales.

### Feature availability

The following table shows features that vary by plan. Features not listed here are available on all plans without restrictions.

| Feature | Starter | Pro | Enterprise |
| - | - | - | - |
| Custom dashboards (Build your own dashboards and custom charts to track metrics specific to your use case. The built-in Cost and quality dashboard is available on all plans.) | ✗ | ✓ | ✓ |
| Environments (Separate development, staging, and production data within your projects.) | ✗ | ✓ | ✓ |
| Dataset snapshots (Save named checkpoints of a dataset to preserve state, pin versions for experiments, and promote to environments.) | ✗ | ✓ | ✓ |
| Human review scorers (Configure custom scoring workflows where human reviewers can provide feedback on traces.) | 1 / project | Unlimited | Unlimited |
| Playground annotations (Annotate playground outputs with thumbs up/down reactions and free text comments, then let Loop suggest prompt improvements based on your feedback.) | ✗ | ✓ | ✓ |
| Sandbox evals (Run evals in an isolated cloud sandbox (AWS Lambda or Modal), controlled from Braintrust. Push an execution artifact once and invoke it on demand from the playground.) | ✗ | ✓ | ✓ |
| Built-in permission groups (Built-in permission groups define what users can do in your organization. Starter includes the Owner group only. Pro includes all four built-in groups: Owner, Engineer, Viewer, and All AI Provider Access. Enterprise supports custom permission groups, project-level and object-level permissions, and project groups.) | Owner only | All four built-in groups | Custom |
| Custom retention policies (Configure custom data retention rules based on your compliance and storage needs.) | ✗ | ✗ | ✓ |
| Export automations (Automatically export logs to an Amazon S3 or Google Cloud Storage bucket on a scheduled basis.) | ✗ | ✗ | ✓ |
| Async queries (Run long-running SQL queries in the background, for large exports or queries that time out, and fetch the results in pages.) | ✗ | ✗ | ✓ |
| Audit logging (Track and review administrative actions in your organization, such as permission changes, member management, and API key creation.) | ✗ | ✗ | ✓ |
| SSO (Single sign-on for centralized identity management.) | Google only | Google only | SAML/OIDC providers |
| Domain mappings (Automatically add users from specific email domains to your organization.) | ✗ | ✗ | ✓ |
| Deployment options (SaaS is available on all plans. BYOC and self-hosted deployments, which keep data in your own cloud, require Enterprise.) | SaaS | SaaS | SaaS, BYOC, and self-hosted |
| SOC2 attestation (Independent audit confirming that Braintrust's security controls are operating effectively over time.) | ✗ | ✗ | ✓ |
| DPA (Data Processing Agreement for GDPR compliance requirements to satisfy data processing obligations.) | ✗ | Click-through | Custom |
| BAA (Business Associate Agreement for HIPAA compliance to support handling of protected health information (PHI).) | ✗ | ✗ | ✓ |

> [!NOTE]
>
> **Legacy plans**: If you signed up before March 16, 2026, you may retain access to features you created even if they're not listed as available on your current plan tier. See [Legacy plans FAQ](https://docs.braintrust.dev/admin/billing/faq.md#legacy-plans) for details.

### Support

All plans include community support.

| Feature | Starter | Pro | Enterprise |
| - | - | - | - |
| Email support | Standard | Priority | Priority |
| Slack support | ✗ | ✗ | ✓ |
| Guaranteed SLAs | ✗ | ✗ | ✓ |
| Custom legal terms | ✗ | ✗ | ✓ |

## Model credits

Monthly [model credits](#usage-limits) cover [built-in models](https://docs.braintrust.dev/admin/ai-providers.md#manage-built-in-models), which you can use without providing your own API key, as well as AI features like [Topics](https://docs.braintrust.dev/observe/topics.md). Built-in models are available on every plan. On Starter, your organization needs at least one owner with a work email address, or a payment method on file, to access them. Calling built-in models directly through the gateway requires a Pro or Enterprise plan. See [Requirements](https://docs.braintrust.dev/admin/ai-providers.md#requirements). After your monthly credits are exhausted, the on-demand token rates below apply.

| Feature | Input tokens | Cached input tokens | Output tokens |
| - | - | - | - |
| Topics (Topics discovers and classifies patterns in your logs. Billed on the total input and output tokens consumed across the whole pipeline, after monthly credits are exhausted.) | \$0.06 / MTok | — | \$0.40 / MTok |
| GPT-6 Astra (A multimodal reasoning model served by Braintrust.) | \$10.00 / MTok | \$1.00 / MTok | \$50.00 / MTok |
| GPT-6.1 Sol (A multimodal reasoning model served by Braintrust.) | \$2.00 / MTok | \$0.10 / MTok | \$10.00 / MTok |
| GPT-6 Sol (A multimodal reasoning model served by Braintrust.) | \$2.00 / MTok | \$0.20 / MTok | \$10.00 / MTok |
| GPT-6 Luna (A multimodal reasoning model served by Braintrust.) | \$0.10 / MTok | \$0.01 / MTok | \$0.50 / MTok |
| GPT-5.6 Sol (A multimodal reasoning model served by Braintrust.) | \$4.00 / MTok | \$0.40 / MTok | \$20.00 / MTok |
| GPT-5.6 Terra (A multimodal reasoning model served by Braintrust.) | \$2.00 / MTok | \$0.20 / MTok | \$12.00 / MTok |
| GPT-5.6 Luna (A multimodal reasoning model served by Braintrust.) | \$0.20 / MTok | \$0.02 / MTok | \$1.20 / MTok |
| GLM-5.2 (An open-source reasoning model served by Braintrust.) | \$1.40 / MTok | \$0.26 / MTok | \$4.40 / MTok |
| GLM-5.3 Flash (An open-source multimodal reasoning model served by Braintrust.) | \$0.15 / MTok | \$0.03 / MTok | \$0.50 / MTok |
| Kimi K3 (An open-source reasoning model served by Braintrust.) | \$3.00 / MTok | \$0.30 / MTok | \$15.00 / MTok |
| DeepSeek V4 Flash 0731 (An open-source reasoning model served by Braintrust.) | \$0.13 / MTok | \$0.028 / MTok | \$0.26 / MTok |
| DeepSeek V4.1 Flash (An open-source multimodal reasoning model served by Braintrust.) | \$0.15 / MTok | \$0.003 / MTok | \$0.60 / MTok |
| Jev (A TypeSafe structured decision model served by Braintrust for scorers and classifiers.) | \$0.042 / MTok | — | \$0 / MTok |

> [!NOTE]
>
> Model credits do not roll over month-to-month. On Starter, if you exhaust your credits and have not [enabled on-demand usage](https://docs.braintrust.dev/admin/billing/change-plan.md#enable-on-demand-usage), Topics and built-in model usage are paused until the next monthly cycle or until you add a card. Enterprise plans have custom model credits and discounted token rates. [Contact Braintrust](https://www.braintrust.dev/contact) to learn more.

## System limits

These limits apply to all users regardless of plan. They ensure platform stability, prevent timeouts, and maintain performance for everyone. The async query limits apply to the Enterprise plan, where async queries are available.

> [!NOTE]
>
> [Self-hosted customers](https://docs.braintrust.dev/admin/self-hosting.md) can adjust some of these limits. [Contact Braintrust](mailto:support@braintrust.dev) for details.

| Limit | Value |
| - | - |
| Function executions (This rate limit applies to API calls that execute prompts, scorers, tools, and other custom code functions. Exceeding this limit returns an HTTP 429 error response.) | 10,000 operations per 10 seconds |
| Git metadata (Git metadata helps track which code version generated experiment results. Git metadata larger than this is truncated. This affects only the metadata attached to experiments, not your actual data.) | 64 KB maximum |
| SQL query timeout (How long a SQL query can run before it times out. For longer-running reads, use async queries.) | 30 seconds |
| Async query runtime (Maximum cumulative runtime for an async query on Braintrust-hosted deployments. Async queries are available on the Enterprise plan.) | 1 hour |
| Async query result size (Maximum size of the results for a single async query on Braintrust-hosted deployments. A query that exceeds this limit stops and reports a failed status.) | 10 GiB |
| Inline code function timeout (Custom code written directly in Braintrust (prompts, scorers, tools) for simple, self-contained logic without external dependencies.) | 240 seconds (4 minutes) |
| Bundled code function timeout (Custom code uploaded with dependencies and external libraries for more complex functions that require npm packages, pip packages, etc.) | 30 seconds (by default) |
| Attachment URL expiration (Signed URLs for uploading or downloading attachments (images, audio, files) expire after 24 hours. Attachments remain stored permanently; only the access URLs expire.) | 1 day |
| Per-span payload size (Maximum size per span in logged traces. JSON payloads up to this limit are automatically indexed. Beyond 20 MB, use JSONAttachment to avoid errors.) | 20 MB |
| Projects per project group (Maximum number of projects that can belong to a single project group. Grant a permission group access to a project group to apply that access to every project in the group.) | 10,000 |
| Project assignments per save (Maximum number of project assignments, counting additions and removals together, that a single save can apply to a project group. To assign more projects than this, save in batches.) | 1,000 |
| MCP request body size (Maximum size of a single JSON request body sent to the MCP server or the Loop runtime. Requests that exceed this limit return an HTTP 413 error.) | 100 MiB |
| Concurrent active Loop threads (Braintrust-hosted Loop applies this limit separately to each member, service token, and automation within an organization. Starting work in another thread at the limit returns an HTTP 429 error.) | 5 on Starter without on-demand usage, 50 on Starter with on-demand usage, Pro, and Enterprise |
| Loop sandbox lifetime (A Loop sandbox is replaced after this long, including time spent asleep. Your thread and its history are unaffected.) | 8 hours |

## Next steps

* Visit the [Pricing page](https://www.braintrust.dev/pricing) to estimate your monthly costs
* [Change your plan](https://docs.braintrust.dev/admin/billing/change-plan.md) to upgrade or downgrade
* [Monitor usage](https://docs.braintrust.dev/admin/billing/monitor-usage.md) to track costs and set up alerts
* Review the [Billing FAQ](https://docs.braintrust.dev/admin/billing/faq.md) for common questions


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.