exaone-4-5-33b (Reasoning)
Intelligence
30.2
#162/523
Coding
23
#187/429
Math
—
Speed
—
Pricing
—
exaone-4-5-33b (Reasoning) is LG AI Research’s model designed for advanced reasoning tasks. It operates at a speed of 0 tokens per second and is priced at $0 per million tokens, making it accessible for various applications.
When to Use exaone-4-5-33b (Reasoning)
✓ Best For
- Complex reasoning tasks
- Educational applications
- Research and analysis
✗ Not Ideal For
- Real-time processing needs
- High-speed applications
How exaone-4-5-33b (Reasoning) Compares
Intelligence Index · Higher is better
XiaomiGoogleLG AI ResearchZ AI
Benchmark Profile
Coding Index
AlibabaInclusionAILG AI ResearchOpenAI
Intelligence · Coding · Math
Intelligence
Coding
Math
All Benchmark Scores (9)
| Benchmark | Score | |
|---|---|---|
| Intelligence Index | 30.2 | |
| Coding Index | 23 | |
| GPQA | 794% | |
| HLE | 116% | |
| SciCode | 28% | |
| IFBench | 58% | |
| LCR | 49.3% | |
| TerminalBench Hard | 20.5% | |
| Tau2 | 78.1% |
Data: Artificial Analysis · Updated: April 27, 2026
Frequently Asked Questions (15)
When was exaone-4-5-33b (Reasoning) released?
exaone-4-5-33b (Reasoning) was released on April 9, 2026.
Who created exaone-4-5-33b (Reasoning)?
exaone-4-5-33b (Reasoning) was created by LG AI Research.
How intelligent is exaone-4-5-33b (Reasoning)?
exaone-4-5-33b (Reasoning) scores 30 on the Artificial Analysis Intelligence Index, placing it well above average among other open weight models of similar size (median: 15).
How verbose is exaone-4-5-33b (Reasoning)?
When evaluated on the Intelligence Index, exaone-4-5-33b (Reasoning) generated 150M output tokens, which is at the higher end compared to other open weight models of similar size (median: 23M).
Is exaone-4-5-33b (Reasoning) a reasoning model?
Yes, exaone-4-5-33b (Reasoning) is a reasoning model. It uses extended thinking or chain-of-thought reasoning to work through complex problems before providing an answer.
What input modalities does exaone-4-5-33b (Reasoning) support?
exaone-4-5-33b (Reasoning) supports text and image input.
What output modalities does exaone-4-5-33b (Reasoning) support?
exaone-4-5-33b (Reasoning) supports text output.
Can exaone-4-5-33b (Reasoning) process images?
Yes, exaone-4-5-33b (Reasoning) supports image input and can analyze, describe, and answer questions about images.
Is exaone-4-5-33b (Reasoning) multimodal?
Yes, exaone-4-5-33b (Reasoning) is multimodal. It can process text and image input and generate text output.
What is the context window of exaone-4-5-33b (Reasoning)?
exaone-4-5-33b (Reasoning) has a context window of 260k tokens. This determines how much text and conversation history the model can process in a single request.
Is exaone-4-5-33b (Reasoning) open source?
Yes, exaone-4-5-33b (Reasoning) is open weights. The model weights are publicly available and can be downloaded for self-hosting.
How many parameters does exaone-4-5-33b (Reasoning) have?
exaone-4-5-33b (Reasoning) has 34.4 billion parameters.
What is the license for exaone-4-5-33b (Reasoning)?
exaone-4-5-33b (Reasoning) is released under the EXAONE AI Model License Agreement 1.2 - NC license. Commercial use requires a separate license agreement.
How does exaone-4-5-33b (Reasoning) perform on benchmarks?
exaone-4-5-33b (Reasoning) achieves a score of 30 on the Artificial Analysis Intelligence Index. This composite benchmark evaluates models across reasoning, knowledge, mathematics, and coding.
Is exaone-4-5-33b (Reasoning) available via API?
exaone-4-5-33b (Reasoning) is an open weights model that can be self-hosted.
