o3 Mini High
OpenAI o3-mini-high is the same model as [o3-mini](/openai/o3-mini) with reasoning_effort set to high. It is not in the StickyPrompts model selector today. The specs are here so you can compare it with the models that are. If your team needs it, tell us.
- Context window
- 200K tokens
- Max output
- 100K tokens
- Input price
- $1.1 / 1M tokens
- Output price
- $4.4 / 1M tokens
- Cached input
- $0.55 / 1M tokens
- Released
- 12 Feb 2025
Capabilities, as the provider publishes them.
Taken from the provider's own model metadata rather than our impressions of it. What we do not publish is a quality or speed score - those are measurements, and we do not run them.
- Tool & function calling
- Extended reasoning
- Structured outputs
- Image input
- Audio input
- Video input
- File input
- Image output
Output tokens cost 4x input on o3 Mini High. Worth knowing before you give it a long-running agent job. StickyPrompts meters both, and Usage Statistics breaks the spend down by model and by user.
The model changes. The workspace around it does not.
New models ship every few weeks. When your prompts, knowledge and rules live in one workspace, moving to a better model is a choice in a dropdown rather than a migration.
- One login, one prompt library and one usage record across every model
- Knowledge bases, agents and connected tools work the same whichever model answers
- Admins allow, block or limit each provider per team in Model & Provider Access
- Sensitive data is warned about, redacted or blocked before it reaches any model
- Optional EU-only inference and zero-data-retention constraints, applied to the whole catalogue
- Compare two to five models on the same prompt before you settle on one
The rest of this provider's line-up.
Other checkpoints from the same maker, with the ones in StickyPrompts marked.
Specifications synced 29 Sept 2026 from the OpenRouter public model catalogue; availability in StickyPrompts checked 29 Sept 2026. Providers change prices and deprecate checkpoints without much notice, so check the provider's own documentation before making a commercial decision on these figures.
Give every team the right model.
Start free with a $5 trial balance. One workspace for every provider, and nothing to migrate when the next model lands.