Supported AI Models
LLMTrader integrates leading language models from multiple providers, giving you the flexibility to choose the AI that best fits your trading strategy. Each model brings a distinct approach to market analysis, risk assessment, and trade execution.
Model Tiers
Every model on the platform falls into one of three tiers. Tiers reflect the general capability and resource intensity of each model:
Premium Tier
Premium models represent the most advanced offerings from each provider. They tend to excel at complex reasoning. Premium models are best suited for strategies that require deep analytical thinking.
Standard Tier
Standard models offer a strong balance of capability and efficiency. They handle most trading strategies well and respond quickly, making them a reliable default choice for the majority of use cases. If you are unsure where to start, a standard-tier model is usually a good pick.
Economy Tier
Economy models prioritize efficiency. They are faster and less resource-intensive, making them ideal for simpler strategies, high-frequency testing, or situations where you want to run many experiments faster.
Available Models
Claude (Anthropic)
Tiers available: Premium, Standard, Economy
Claude models from Anthropic are known for advanced reasoning, careful instruction-following, and nuanced analysis. Claude tends to produce well-structured trading rationale and is particularly strong at interpreting complex, multi-condition prompts. If your strategy involves detailed risk rules, layered conditions, or requires the AI to explain its reasoning clearly, Claude is an excellent choice.
Models:
- Claude Opus 4.6 (Premium): Maximum reasoning depth and analytical power
- Claude Sonnet 4.6 (Standard): Strong balance of capability and speed
- Claude Haiku 4.5 (Economy): Fast and efficient for high-frequency testing
Strengths:
- Advanced multi-step reasoning
- Precise prompt interpretation
- Thorough risk analysis
- Clear, detailed trade rationale
Gemini (Google)
Tiers available: Premium, Standard
Gemini models from Google are known for handling large amounts of context efficiently and delivering fast responses. Gemini performs well when your strategy requires the AI to consider extensive market data simultaneously or when you need quick decision-making. It is also strong at synthesizing information from multiple sources into coherent trading logic.
Models:
- Gemini 3 Pro (Premium): Deep analysis with massive context window
- Gemini 3 Flash (Standard): Fastest response times on the platform
Strengths:
- Fast response times
- Efficient processing of large data sets
- Strong information synthesis
- Good performance on data-heavy strategies
Kimi (Moonshot)
Tiers available: Standard
Kimi by Moonshot specializes in long-context understanding, making it well-suited for strategies that reference extensive market history, detailed rule sets, or complex multi-asset analysis. If your prompt is lengthy or your strategy requires the AI to keep track of many variables simultaneously, Kimi handles this particularly well.
Models:
- Kimi K2.5 (Standard): Long-context specialist with strong analytical reasoning
Strengths:
- Long-context comprehension
- Handling complex, detailed prompts
- Multi-variable tracking
- Consistent performance on lengthy instructions
DeepSeek
Tiers available: Standard, Economy
DeepSeek models are recognized for strong reasoning capabilities combined with cost-effectiveness. They offer solid analytical performance, making them a practical choice for users who want to run multiple strategy tests or need reliable reasoning without committing to premium-tier resources. DeepSeek is especially popular for systematic, logic-driven strategies.
Models:
- DeepSeek Reasoner (Standard): Advanced chain-of-thought reasoning for complex strategies
- DeepSeek Chat (Economy): Fast and cost-effective for bulk testing
Strengths:
- Strong logical reasoning
- Cost-effective for bulk testing
- Reliable systematic analysis
- Good performance on rule-based strategies
Qwen (Alibaba)
Tiers available: Standard, Economy
Qwen models from Alibaba bring strong multilingual capabilities and solid general reasoning. They perform well across a variety of strategy types and are a good choice for users who write prompts in languages other than English or who want a well-rounded model for diverse trading approaches.
Models:
- Qwen Max (Standard): Strongest reasoning in the Qwen family
- Qwen Plus (Economy): Accessible option for diverse strategies
Strengths:
- Multilingual prompt support
- Versatile across strategy types
- Solid general-purpose reasoning
- Accessible economy-tier option
Choosing a Model
There is no single "best" model. The right choice depends on your strategy, goals, and preferences. Here are some guidelines:
| If you want... | Consider... |
|---|---|
| Maximum reasoning depth | Claude Opus 4.6, Gemini 3 Pro, DeepSeek Reasoner |
| Fast responses | Gemini 3 Flash, Claude Haiku 4.5 |
| Complex, lengthy prompts | Kimi K2.5 |
| Bulk testing on a budget | DeepSeek Chat, Qwen Plus, Claude Haiku 4.5 |
| Non-English prompts | Qwen Max, Qwen Plus |
| Balanced performance | Claude Sonnet 4.6, Gemini 3 Flash, Kimi K2.5 |
Try Multi-Model Testing
The most reliable way to find the right model for your strategy is to test it across multiple models simultaneously. LLMTrader's Alpha Testing feature makes this easy. Run the same prompt on several models and compare their results directly.
Model Updates
The AI landscape evolves rapidly. LLMTrader regularly evaluates and integrates new models as they become available. When a new model is added or an existing model is upgraded, it will be announced on the platform and reflected in this documentation.
Models may also be retired if they are superseded by significantly improved versions from the same provider. In such cases, users will be notified in advance and given time to transition their strategies.
Important Notes
No model can predict the future. Language models analyze patterns, follow instructions, and reason about market conditions, but they have no special ability to forecast prices. Their value lies in consistent, disciplined execution of your strategy, not in generating guaranteed returns.
Model behavior varies. The same model may produce different trading decisions when run multiple times on the same prompt. This is an inherent characteristic of language models, not a bug. Multi-model testing helps smooth out this variability.
Your prompt matters most. The quality of your strategy prompt has a far greater impact on results than model selection alone. A well-crafted prompt on a standard-tier model will typically outperform a vague prompt on a premium-tier model. See Alpha Testing for prompt-writing guidance.
Disclaimer: The availability and performance of AI models may change over time as providers update their offerings. Model descriptions reflect general characteristics and are not guarantees of specific behavior. AI-generated trading decisions are experimental and should not be treated as financial advice.