ERNIE 4.0: pricing, context window and benchmarks
ERNIE 4.0 is a multimodal model from Baidu. It offers a 32,768-token context window, costs $11.37 per million input tokens and $45.44 per million output tokens, and averages 76.7 across the index's benchmark suite.
Pricing
ERNIE 4.0 costs $11.37 per million input tokens and $45.44 per million output tokens. A blended 3:1 input-to-output mix works out to about $19.89 per million tokens. Cached input is listed at $1.68 per million tokens.
| Item | Price | Notes |
|---|---|---|
| Input | $11.37 | per 1M tokens |
| Output | $45.44 | per 1M tokens |
| Cached input | $1.68 | per 1M tokens |
| Blended 3:1 | $19.89 | estimate per 1M tokens |
| 1M in + 1M out | $56.81 | 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 | 79.0 | 208 | 65th | |
| GPQA Diamond | 74.6 | 151 | 75th | |
| SWE-bench Verified | 66.9 | 170 | 72th | |
| HumanEval | 82.5 | 174 | 71th | |
| MATH-500 | 82.1 | 162 | 73th | |
| AIME 2025 | 73.7 | 191 | 68th | |
| LiveCodeBench | 66.9 | 253 | 58th | |
| IFEval | 82.8 | 248 | 59th | |
| TruthfulQA | 77.0 | 280 | 53th | |
| HellaSwag | 87.9 | 144 | 76th | |
| MMMU | 70.2 | 290 | 52th | |
| Arena Elo | 1390.2 | 249 | 59th |
What ERNIE 4.0 is good for
ERNIE 4.0 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 $11.37 per million input tokens, it sits in the mid-range of the index, and the 32,768-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 $11.37 per million tokens.
How it compares
| Model | Provider | Context | Input / 1M | Avg score |
|---|---|---|---|---|
| Llama 3.1 Turbo | Together AI | 200000 | $4.17 | 76.8 |
| Llama 3.3 Groq | Groq | 32768 | $3.98 | 76.8 |
| Magistral Chat | Mistral AI | 128000 | $4.33 | 76.8 |
| ERNIE 4.0 | Baidu | 32768 | $11.37 | 76.7 |
| Llama 3.1 Turbo Instruct | Together AI | 32768 | $4.21 | 76.7 |
| Llama 3.3 Groq Instruct | Groq | 65536 | $4.08 | 76.7 |
FAQ
How much does ERNIE 4.0 cost?
ERNIE 4.0 is listed at $11.37 per million input tokens and $45.44 per million output tokens. Cached input is $1.68 per million tokens. A blended 3:1 input-to-output mix is about $19.89 per million tokens.
What is the context window of ERNIE 4.0?
ERNIE 4.0 has a context window of 32,768 tokens and can emit up to 32,768 output tokens per request in this index.
Is ERNIE 4.0 open weights?
No. ERNIE 4.0 is a proprietary model with a proprietary licence, so you cannot download the weights.
How does ERNIE 4.0 score on benchmarks?
ERNIE 4.0 averages 76.7 across the scored benchmarks in this index, ranking #200 of 600 models. Its strongest relative result is Arena Elo at 1390.2.
What is ERNIE 4.0 best for?
ERNIE 4.0 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 $11.37 per million input tokens, it sits in the mid-range of the index, and the 32,768-token context window is enough for large files and multi-file diffs.