Compare models
Pick two to four models. No markup on model usage: you pay the provider's list price.
| Attribute | Lion |   GPT-6 Astra |
|---|---|---|
| Provider | Safeguard | OpenAI |
| Model id | safeguard/lion | openai/gpt-6-astra |
| Status | Available via Safeguard | Available via Safeguard |
| Input / 1M tokens | $0.000001 | $10 |
| Output / 1M tokens | $0.000001 | $50 |
| Price notes | Standard tier, prompts <=272K input tokens. Over 272K: $20 in / $75 out. Cached input $1.00. Batch and Flex 50% off. | |
| Context | 128K | 1.05M |
| Parameters | 15B | Not published |
| Max output | Not published | 128,000 tokens |
| Modality | Text | Text and image |
| Weights | Safeguard | Closed |
| Zero data retention | Not stated | Available |
| Prompts go to | Safeguard, on its own infrastructure | OpenAI (no processing-region constraint by default) |
| Retention | Customer data is never used to train shared models. | API data not used for training unless you opt in, abuse monitoring logs that can include prompts and responses kept up to 30 days by default, some features store application state |
| Region | The region your Safeguard workspace runs in | Global by default, data residency projects in US, EU and other regions (approval required, 10% uplift for models released on or after 2026-03-05) |
| Price checked | 2026-10-09 | 2026-10-10 |
Self-healing security runs on Safeguard.
Your first fix PR is minutes away.
No sales call required, even your agent can complete the purchase over MCP.