Overview
OpenAI provides the GPT family of models, including the latest GPT-5 series and specialized reasoning models. These models offer industry-leading performance across a wide range of tasks from simple chat to complex reasoning.Available Models
GPT-5 Series (Latest)
GPT-5
5 credits ” Flagship model with exceptional reasoning and creativity
- 200K context window
- Best for: Complex tasks, creative writing, advanced reasoning
- Speed: Medium ” Cost: High
GPT-5 Mini
3 credits ” Balanced performance and speed
- 200K context window
- Best for: Everyday tasks with strong reasoning
- Speed: Fast ” Cost: Medium
GPT-5 Nano
2 credits ” Ultra-lightweight and fast
- 128K context window
- Best for: Simple tasks, high-volume applications
- Speed: Very Fast ” Cost: Low
GPT-5.1
4 credits ” Enhanced reasoning model
- 200K context window
- Reasoning model with extended thinking
- Speed: Medium ” Cost: High
O-Series Reasoning Models
Reasoning models use explicit step-by-step thinking for complex problem-solving. They require higher minimum completion tokens (2000-3000) and are best for analytical tasks.
| Model | Credits | Context | Min Tokens | Best For |
|---|---|---|---|---|
| O1 | 6 | 200K | 3000 | Deep reasoning, complex problems |
| O1 Mini | 3 | 128K | 2000 | Faster reasoning with strong analytics |
| O3 Mini | 3 | 128K | 2000 | Next-gen compact reasoning |
GPT-4 Series
| Model | Credits | Context | Description |
|---|---|---|---|
| GPT-4o | 5 | 128K | Specialized for specific tasks, excellent performance |
| GPT-4o Mini | 1 | 128K | Balanced for everyday tasks, efficient |
| GPT-4.1 Nano | 0.5 | 64K | Lightweight for speed and low-cost |
| GPT-4.1 Mini | 1 | 128K | Efficient with solid everyday performance |
Setup
Using BoostGPT-Hosted API Keys
1
Select OpenAI Models
In your BoostGPT dashboard, simply select any OpenAI model when creating or configuring your bot. No API key needed!
2
Choose Your Model
Select based on your needs:
- GPT-5: Complex reasoning, creative tasks
- GPT-5 Mini: Balanced everyday use
- O1/O3 Mini: Deep analytical work
- GPT-4o Mini: Cost-effective simple tasks
Using Your Own OpenAI API Key
Want to use your own OpenAI API key? See the Bring Your Own Keys guide.
- Dashboard Setup
- Core SDK
- Router SDK
1
Navigate to Integrations
Go to app.boostgpt.co and select Integrations
2
Select OpenAI
Find and click on the OpenAI provider
3
Add API Key
Enter your OpenAI API key and select which agents will use this key
4
Save Configuration
Click save to apply your custom API key
Model Selection Guide
When to Use Each Model
GPT-5 - Complex & Creative Tasks
GPT-5 - Complex & Creative Tasks
Best for:
- Creative writing and content generation
- Complex reasoning and analysis
- Multi-step problem solving
- High-stakes customer interactions
- Simple FAQ responses
- High-volume/cost-sensitive applications
GPT-5 Mini - Everyday Workhorse
GPT-5 Mini - Everyday Workhorse
Best for:
- Customer support chatbots
- General conversation
- Content moderation
- Most production use cases
GPT-5 Nano - Speed & Scale
GPT-5 Nano - Speed & Scale
Best for:
- High-volume applications
- Simple queries and responses
- Quick classifications
- Prototyping and testing
O1/O3 Mini - Deep Reasoning
O1/O3 Mini - Deep Reasoning
Best for:
- Mathematical problem solving
- Code debugging and optimization
- Scientific analysis
- Strategic planning
GPT-4o Mini - Budget Option
GPT-4o Mini - Budget Option
Best for:
- Development and testing
- Simple chatbots
- Low-budget projects
- Learning and experimentation
Troubleshooting
Rate limit errors (429)
Rate limit errors (429)
Cause: Too many requests to OpenAI APISolutions:
- Implement request throttling
- Use exponential backoff retries
- Upgrade your OpenAI API tier
- Switch to GPT-4o Mini for high-volume use
Context length exceeded
Context length exceeded
Cause: Input + completion tokens exceed model’s context windowSolutions:
- Truncate conversation history
- Use models with larger context (GPT-5: 200K)
- Implement sliding window for message history
Slow responses with O1 models
Slow responses with O1 models
Expected behavior: Reasoning models take longerSolutions:
- Set user expectations (show “thinking…” indicator)
- Use O1 Mini instead of O1 for faster responses
- Reserve reasoning models for truly complex tasks
High costs
High costs
Solutions:
- Use GPT-5 Mini or GPT-4o Mini for most tasks
- Implement dynamic model selection
- Cache common responses
- Set max_tokens to limit completion length