Model comparison
DeepSeek R1 vs GPT-5.4 Mini
DeepSeek
DeepSeek R1
The open-weights reasoning model that reset the cost curve.
OpenAI
GPT-5.4 Mini
GPT-5 economics for high-volume routine tasks.
Specs
| Metric | DeepSeek R1 | GPT-5.4 Mini |
|---|---|---|
| Context window | 164K tokens | 400K tokens↑ |
| Input $/1M tokens | $0.700↑ | $0.750 |
| Output $/1M tokens | $2.50↑ | $4.50 |
| Modalities | Text | File · Image · Text |
| Open weights | Yes | No |
| Released | Jan 2025 | Mar 2026 |
Capability differences
| Capability | DeepSeek R1 | GPT-5.4 Mini |
|---|---|---|
| Tool use | No | Yes |
| Vision | No | Yes |
| Extended thinking | Yes | No |
| Prompt caching | No | Yes |
| Open weights | Yes | No |
Key differences
- -DeepSeek R1 offers 164K context; GPT-5.4 Mini offers 400K.
- -Input pricing: DeepSeek R1 at $0.700/1M vs GPT-5.4 Mini at $0.750/1M.
- -DeepSeek R1 is open weights; GPT-5.4 Mini is proprietary.