token.app › Compare › GPT-3.5 Turbo vs Phi 4
GPT-3.5 Turbo vs Phi 4
Phi 4 leads on 2 of 2 shared benchmarks. Phi 4 is 8.6× cheaper per token on a blended 3:1 basis. GPT-3.5 Turbo accepts the larger context window (16K).
Head to head
Blended price uses a 3:1 input:output mix
| GPT-3.5 Turbo | Phi 4 | |
|---|---|---|
| Provider | OpenAI | Microsoft |
| Input $/1M | $0.500 | $0.070 |
| Output $/1M | $1.50 | $0.140 |
| Blended $/1M | $0.750 | $0.088 |
| Context window | 16K | 16K |
| Max output | 4K | 16K |
| Released | May 2023 | Jan 2025 |
| Licence | Proprietary | Open weights |
| GPQA Diamond | 27.2% | 56.1% |
| OTIS Mock AIME | 2.2% | 13.8% |
Benchmark scores independently run and published by Epoch AI, 'AI Benchmarking Hub', used under CC BY 4.0. Only benchmarks with a score for BOTH models are compared.
Choose GPT-3.5 Turbo if…
- Larger context window (16K)
Choose Phi 4 if…
- Costs less per token — $0.088 vs $0.750 blended
- Higher GPQA Diamond (56.1% vs 27.2%)
- Higher OTIS Mock AIME (13.8% vs 2.2%)
- Open weights — self-hostable
- Newer model (released Jan 2025)
FAQ
Which is better, GPT-3.5 Turbo or Phi 4?
Across the 2 benchmarks both models have independently-run scores for, GPT-3.5 Turbo leads on 0 and Phi 4 leads on 2. Benchmarks are one input — price and context window are on this page too.
Is GPT-3.5 Turbo cheaper than Phi 4?
Phi 4 is cheaper: $0.088 versus $0.750 per million tokens blended at 3:1 input:output. Input alone: $0.500 vs $0.070. Output alone: $1.50 vs $0.140.
What context windows do GPT-3.5 Turbo and Phi 4 have?
GPT-3.5 Turbo: 16,385 tokens. Phi 4: 16,384 tokens.
Who makes GPT-3.5 Turbo and Phi 4?
GPT-3.5 Turbo is made by OpenAI. Phi 4 is made by Microsoft.