ERNIE-4.5-VL-28B-A3B-Thinking vs deepseek-ai/DeepSeek-V3
deepseek-ai/DeepSeek-V3 is 1.2× cheaper for input tokens: ERNIE-4.5-VL-28B-A3B-Thinking costs $0.39 per M and deepseek-ai/DeepSeek-V3 costs $0.32 per M. ERNIE-4.5-VL-28B-A3B-Thinking is 2.3× cheaper for output tokens: ERNIE-4.5-VL-28B-A3B-Thinking costs $0.39 per M and deepseek-ai/DeepSeek-V3 costs $0.89 per M. deepseek-ai/DeepSeek-V3 reads more: ERNIE-4.5-VL-28B-A3B-Thinking takes up to 131,072 tokens of context and deepseek-ai/DeepSeek-V3 up to 163,840. ERNIE-4.5-VL-28B-A3B-Thinking accepts image, text; deepseek-ai/DeepSeek-V3 accepts text.
| ERNIE-4.5-VL-28B-A3B-Thinking | deepseek-ai/DeepSeek-V3 | |
|---|---|---|
| Model id | baidu/ernie-4.5-vl-28b-a3b-thinking | deepseek-ai/DeepSeek-V3 |
| Input | $0.39 per M | $0.32 per M |
| Output | $0.39 per M | $0.89 per M |
| Context length | 131,072 tokens | 163,840 tokens |
| Accepts | image, text | text |
| Produces | text | text |
| Served by | novita | deepinfra |
Prices are the providers' list prices per million tokens, as of . Both run on your own provider key through one OpenAI-compatible endpoint, with no fee on the tokens.