Baidu model product
ERNIE 4.5
5 is a 21B parameter LLM from Baidu with improved reasoning, instruction following, and performance in mathematics, coding, and creative writing.
Updated Sep 8, 2026. Default version: ERNIE 4.5
LLMBoard Score0.0ERNIE 4.5
Coverage60%20 benchmark families
Context window128KTokens
Official input priceN/AOfficial price unavailable
ERNIE 4.5 Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
ERNIE 4.5 Benchmark Results
Benchmark scores for ERNIE 4.5.
Sort by
Benchmark | Score | Rank | Participants | Percentile | Evidence | Evaluated |
|---|
| BenchmarkCLUEWSC | Score48.60% | Rank03 | Participants3 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkMBPP+ | Score40.20% | Rank04 | Participants4 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkCMMLU | Score39.80% | Rank06 | Participants6 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkARC-E | Score60.70% | Rank08 | Participants8 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkAGIEval | Score28.50% | Rank10 | Participants10 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkHumanEval+ | Score25.00% | Rank10 | Participants10 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkPIQA | Score55.20% | Rank11 | Participants11 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkBBH | Score30.40% | Rank12 | Participants12 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkC-Eval | Score40.70% | Rank18 | Participants18 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkWinogrande | Score51.30% | Rank22 | Participants22 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkHellaSwag | Score33.00% | Rank27 | Participants27 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkDROP | Score28.60% | Rank30 | Participants30 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkARC-C | Score40.60% | Rank34 | Participants34 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkSimpleQA | Score1.80% | Rank47 | Participants47 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkGSM8k | Score25.20% | Rank48 | Participants48 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkMMLU-Redux | Score43.20% | Rank48 | Participants48 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkMATH | Score12.40% | Rank71 | Participants71 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkMMLU | Score41.90% | Rank101 | Participants101 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkGPQA | Score74.00% | Rank123 | Participants247 | Percentile50.41% | EvidenceC | Evaluated |
| BenchmarkMMLU-Pro | Score16.00% | Rank137 | Participants138 | Percentile0.73% | EvidenceC | Evaluated |
ERNIE 4.5 Arena Results
Preference and agent-evaluation results for the default version.
No Arena results
The default version does not have a matching Arena result yet.
ERNIE 4.5 Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- N/A
- Official provider
- N/A
- Lowest third-party
- N/A
- Tracked offerings
- 0
No provider prices
The default version has no current input or output token prices.
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
ERNIE 4.5 Runtime Performance
Provider-specific output speed and catalog latency for ERNIE 4.5. Runtime does not affect the capability score.
No runtime data
No provider-specific speed or latency record is linked to the default version yet.
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
ERNIE 4.5 Specifications
Technical details for the model's default version.
- Version
- ERNIE 4.5
- Released
- Jun 25, 2025
- Knowledge cutoff
- Unknown
- Parameters
- 21B
- Context window
- 128K
- Max output
- 65.5K
- Inputs
- text
- Outputs
- text
- Open weights
- No
- License
- Proprietary
ERNIE 4.5 Versions
Available versions of this model. The score column identifies the version used in the overall ranking.
Sort by
Version | Released | LLMBoard | Parameters | Context | Max output | Open weights | License |
|---|
| VersionERNIE 4.5 | Released | LLMBoard0.00 | Parameters21B | Context128K | Max output65.5K | Open weightsNo | LicenseProprietary |
Models similar to ERNIE 4.5
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
BA
ERNIE 5.0
Baidu
OP
GPT-3.5-turbo
OpenAI
OP
GPT-5.1-Codex-Mini
OpenAI
IB
IBM Granite 4.0 Tiny
IBM
XA
Grok 1.5
xAI
XA
Grok 1.5V
xAI
What is ERNIE 4.5?
Key information about ERNIE 4.5 and its available data.
0 with advanced knowledge integration, multilingual support, and optimized thinking capabilities. It uses A3B quantization for efficient inference.
Data as of 2026-09-08.
FAQ
Common questions about ERNIE 4.5.
When was ERNIE 4.5 released?
ERNIE 4.5's default version was released on Jun 25, 2025.
How much does ERNIE 4.5 cost?
No official standard PAYG price is currently available for ERNIE 4.5.
Who created ERNIE 4.5?
ERNIE 4.5 was created by Baidu.
What is the context window for ERNIE 4.5?
The default version has a 128K token context window.
Is ERNIE 4.5 open weight?
No. The default version is not marked as having publicly available weights.
How many API providers offer ERNIE 4.5?
No provider offering is currently linked to the default version.