Blackbox Priority Service Tier
Service · Managed service by Blackbox AI
Trade off cost for lower latency and higher availability on frontier models.
A managed performance service for Blackbox enterprise customers to access higher availability, faster time-to-first-token, and lower latency for frontier models.
- Languages
- English
Category: AI Infrastructure
- Deployment
- Cloud API (enterprise.blackbox.ai)
- Engagement model
- Metered per-token pricing based on the tier actually served
- Delivery
- Online via OpenAI-compatible API endpoint
- Request Parameter
- service_tier: "priority" or "default"
- Interface
- OpenAI-compatible API (/v1/chat/completions)
- Protocol
- HTTPS with Bearer Token Authorization
- Higher availability and faster processing for frontier models
- Faster time-to-first-token (TTFT) and lower E2E latency
- Automatic best-effort routing with fallback to standard rates
- OpenAI-compatible API integration via service_tier field
Overview
Blackbox Priority Service Tier is a managed performance service for enterprise teams that need to trade off cost for lower latency and higher availability. It allows users to request prioritized processing for specific frontier models, including GPT-5.5 and the GPT-5.6 family (Sol, Luna, and Terra). This ensures that production workloads can optimize for speed without risking request failure.
Performance Optimization
- Faster TTFT: The priority tier delivers a faster time-to-first-token (TTFT) compared to the standard average.
- Lower E2E Latency: Requests served on the priority tier benefit from lower end-to-end (E2E) latency.
- Higher Output Throughput: The priority tier provides higher output throughput measured in tokens per second.
- Higher Availability: Priority processing offers higher availability for mission-critical workloads.
Service Management
- API Request Parameter: Users can request the priority tier by passing the service_tier field in the request body of a chat completion.
- Applied Tier Validation: The tier actually served is returned in the response as the service_tier field to confirm billing.
- Best-effort Routing: If priority capacity is full, the request is automatically downgraded to the default tier and billed at the standard rate.
- Streaming Support: Service tiers are fully supported for streaming completions via Server-Sent Events.
What's included
- Higher availability, faster processing, and lower latency for supported frontier models
Deliverables
- Prioritized API request processing, performance metrics (TTFT, E2E latency, throughput), and billing confirmation via response fields
How it works
- Pass service_tier: priority in API request body; 2. System checks capacity; 3. System serves request at priority or downgrades to default; 4. Response returns applied tier; 5. Billing adjusted based on served tier
Published results
- Faster time-to-first-token (TTFT); Lower end-to-end (E2E) latency; Higher output throughput (tokens per second)
Use cases
- Enterprise developers reduce streaming latency for user-facing applications by requesting priority processing for GPT-5.6 models.
- Infrastructure teams ensure higher availability for mission-critical workloads during peak traffic periods using best-effort priority routing.
- Data engineers optimize throughput for high-volume inference tasks while maintaining a fallback to standard rates when capacity is full.
Specifications
- Deployment
- Cloud API (enterprise.blackbox.ai)
- Engagement model
- Metered per-token pricing based on the tier actually served
- Delivery
- Online via OpenAI-compatible API endpoint
- Prerequisites
- Blackbox API Key and use of a supported model (GPT-5.5 or GPT-5.6 variants)
- Supported Models
- openai/gpt-5.5, openai/gpt-5.6-sol, openai/gpt-5.6-luna, openai/gpt-5.6-terra
- Request Parameter
- service_tier: "priority" or "default"
- Interface
- OpenAI-compatible API (/v1/chat/completions)
- Protocol
- HTTPS with Bearer Token Authorization
- Streaming Support
- Server-Sent Events (SSE) supported
- Billing Metric
- Per-token rates based on served tier
How it works
- Request Priority Tier: Include the service_tier parameter set to "priority" in the JSON body of your chat completion request to the Blackbox enterprise endpoint.
- Automated Routing Decision: The system evaluates capacity and model support; if granted, the request is processed with higher availability and speed.
- Verify Applied Tier: Check the service_tier field in the API response to confirm if the request was served at priority or downgraded to default.
Added by AntSpace from public sources (docs.blackbox.ai).