One agent session
1M input tokens and 100K output tokens, no caching.
- Input1M x $0.84 = $0.84
- Output0.1M x $3.38 = $0.338
$1.18per session
Baidu - Closed weights
ERNIE 5.0 is a closed-weights model from Baidu that costs $0.84 per 1M input tokens and $3.38 per 1M output tokens on the first-party API, with a 128K-token context window.
Last verified from intl.cloud.baidu.com. All sources
| Item | Price / 1M tokens |
|---|---|
| InputPrompts up to 32,000 tokens | $0.84 |
| OutputPrompts up to 32,000 tokens | $3.38 |
| Long-context inputPrompts over 32,000 tokens | $1.40 |
| Long-context outputPrompts over 32,000 tokens | $5.60 |
Note: Baidu AI Cloud international (Qianfan) pricing, page updated 2026-09-07. Text Generation table: Input<=32k is $0.84 in / $3.38 out; 32k<Input<=128k is $1.40 in / $5.60 out (recorded as aboveTokens 32000). The same page's Visual Understanding and Deep Thinking tables list ERNIE 5.0 at a flat $1.40 in / $5.60 out with no prompt-length tier, so image input and thinking use bill at the higher rate. No cache-hit or batch price listed. Context length stated as '128k' (max input 119k), recorded as 128000; max output [1, 65536]; Deep Thinking chain-of-thought length 60k, and max_tokens caps reasoning_content plus content together. The release date is left blank: Baidu's own press release (Nov 13, 2025) launched only a preview of ERNIE 5.0 and the Feb 6, 2026 ERNIE blog post is a technical introduction, so no primary page read states the date the priced non-preview model went on sale. Not listed in the Vercel AI Gateway model list on 2026-10-17.
ernie-5.0Worked from the standard list rates above, with the math shown so you can swap in your own token counts. These use the base rates, which apply while each request's prompt stays at or under 32,000 tokens; above that, ERNIE 5.0 bills $1.40 input and $5.60 output for the whole request.
1M input tokens and 100K output tokens, no caching.
$1.18per session
10,000 requests a day at 2K input and 500 output tokens each.
$33.70per day
$1,011.00 per 30-day month
| Model | Input / 1M | Output / 1M |
|---|---|---|
| ERNIE 5.0 (this page) | $0.84 | $3.38 |
| Amazon Nova ProAmazon | $0.80 | $3.20 |
| GLM-5Z.ai | $1 | $3.20 |
| Qwen3.5-122B-A10BAlibaba Cloud (Qwen) | $0.40 | $3.20 |
| Qwen3.5-397B-A17BAlibaba Cloud (Qwen) | $0.60 | $3.60 |
| ERNIE 5.1Baidu | $0.56 | $2.53 |
| ERNIE 4.5 Turbo VLBaidu | $0.42 | $1.27 |
| ERNIE 4.5 Turbo VL (32K)Baidu | $0.42 | $1.27 |
| ERNIE 4.5 Turbo (128K)Baidu | $0.11 | $0.45 |
| ERNIE 4.5 Turbo (32K)Baidu | $0.11 | $0.45 |
ERNIE 5.0 costs $0.84 per 1M input tokens and $3.38 per 1M output tokens on Baidu's first-party API, per intl.cloud.baidu.com on October 17, 2026. Prompts over 32,000 tokens are billed at $1.40 input and $5.60 output for the whole request.
ERNIE 5.0 has a 128,000-token context window and returns up to 65,536 output tokens per response.
No. ERNIE 5.0 is a closed-weights model. You call it through Baidu's API as ernie-5.0.
Last verified from intl.cloud.baidu.com.
Newsletter
One email per week. New model releases, price cuts and what they mean for your bill.