Overview
GPT-4 Turbo was the previous flagship model, before the release of GPT-4o. Although it still provides strong performance, generally, GPT-4o is preferred.
Key Specifications
- Context Window: 128,000 tokens
- Output Limit: 4,096 tokens
- Training Cutoff: December 2023
- Pricing:
- Input: $10.00 per million tokens
- Output: $30.00 per million tokens
Performance Highlights
- Vision Capabilities: Can analyze and reason about images
- Complex Reasoning: Excels at multi-step problem solving
- Function Calling: Enhanced API capabilities including JSON mode
- Consistency: High reliability across diverse task types
Benchmark Results
Strong performance across our benchmarks:
- TaxEval: High accuracy in tax reasoning tasks
- LegalBench: Superior performance in legal analysis
- ContractLaw: Excellent at contract interpretation
- CaseLaw: Strong understanding of legal precedents
Use Case Recommendations
Best suited for:
- Complex legal analysis
- Vision-based tasks
Limitations
- No longer the most recent model from OpenAI - newer models are preferred.
- Shorter output limit compared to some competitors
- Higher costs for similar performance to GPT-4o
Comparison with Other Models
- Generally, less performance than GPT-4o at a higher cost.
- Comparable performance to Claude 3 Opus
- Higher cost than Gemini Pro 1.5
- Better vision capabilities than most competitors