ERNIE 4.5: pricing, context window and benchmarks
ERNIE 4.5 is a multimodal model from Baidu. It offers a 65,536-token context window, costs $13.44 per million input tokens and $56.59 per million output tokens, and averages 80.0 across the index's benchmark suite.
Pricing
ERNIE 4.5 costs $13.44 per million input tokens and $56.59 per million output tokens. A blended 3:1 input-to-output mix works out to about $24.23 per million tokens. Cached input is listed at $1.99 per million tokens.
| Item | Price | Notes |
|---|---|---|
| Input | $13.44 | per 1M tokens |
| Output | $56.59 | per 1M tokens |
| Cached input | $1.99 | per 1M tokens |
| Blended 3:1 | $24.23 | estimate per 1M tokens |
| 1M in + 1M out | $70.03 | estimate |
Benchmarks
Scores below are modelled estimates on a 0–100 scale. Use them to compare relative position, not as official vendor results.
| Benchmark | Score | Rank | Percentile | Relative |
|---|---|---|---|---|
| MMLU-Pro | 83.7 | 86 | 86th | |
| GPQA Diamond | 77.2 | 102 | 83th | |
| SWE-bench Verified | 68.2 | 140 | 77th | |
| HumanEval | 83.9 | 134 | 78th | |
| MATH-500 | 81.3 | 181 | 70th | |
| AIME 2025 | 75.9 | 142 | 76th | |
| LiveCodeBench | 72.9 | 131 | 78th | |
| IFEval | 86.4 | 146 | 76th | |
| TruthfulQA | 83.0 | 106 | 82th | |
| HellaSwag | 89.2 | 108 | 82th | |
| MMMU | 78.0 | 97 | 84th | |
| Arena Elo | 1414.8 | 168 | 72th |
What ERNIE 4.5 is good for
ERNIE 4.5 is tuned toward coding work. It is a reasonable candidate for code generation, review, debugging and repository-scale tasks, especially when paired with a retrieval layer. At $13.44 per million input tokens, it sits in the mid-range of the index, and the 65,536-token context window is enough for large files and multi-file diffs.
Strengths
- Accepts more than text, including image input where the provider supports it.
- Coding benchmarks are a relative strength within its tier.
Trade-offs
- Proprietary licence: you cannot download the weights or self-host the model.
- Premium input pricing at $13.44 per million tokens.
How it compares
| Model | Provider | Context | Input / 1M | Avg score |
|---|---|---|---|---|
| GLM-4.6 | Zhipu AI | 128000 | $5.10 | 80.2 |
| DeepSeek-R1 Base | DeepSeek | 400000 | $5.05 | 80.1 |
| Claude Haiku 4.5 Turbo | Anthropic | 200000 | $14.37 | 80.0 |
| ERNIE 4.5 | Baidu | 65536 | $13.44 | 80.0 |
| Command A Max | Cohere | 400000 | $14.28 | 79.9 |
| Reka Core Pro | Reka AI | 65536 | $13.44 | 79.9 |
FAQ
How much does ERNIE 4.5 cost?
ERNIE 4.5 is listed at $13.44 per million input tokens and $56.59 per million output tokens. Cached input is $1.99 per million tokens. A blended 3:1 input-to-output mix is about $24.23 per million tokens.
What is the context window of ERNIE 4.5?
ERNIE 4.5 has a context window of 65,536 tokens and can emit up to 32,768 output tokens per request in this index.
Is ERNIE 4.5 open weights?
No. ERNIE 4.5 is a proprietary model with a proprietary licence, so you cannot download the weights.
How does ERNIE 4.5 score on benchmarks?
ERNIE 4.5 averages 80.0 across the scored benchmarks in this index, ranking #120 of 600 models. Its strongest relative result is Arena Elo at 1414.8.
What is ERNIE 4.5 best for?
ERNIE 4.5 is tuned toward coding work. It is a reasonable candidate for code generation, review, debugging and repository-scale tasks, especially when paired with a retrieval layer. At $13.44 per million input tokens, it sits in the mid-range of the index, and the 65,536-token context window is enough for large files and multi-file diffs.