Supported Models

Advanced reasoning, coding, and long-horizon agent workflows

ZAI's latest flagship with strong bilingual reasoning, long-context understanding, and tool use

Native multimodal agentic model with long-context capabilities

Alibaba's largest Qwen3.5 MoE model for complex reasoning

OpenAI open-weight 120B model with transparent weights

Meta's flagship 70B instruction-following model

Strong coding and reasoning performance at low cost

Multimodal flagship with advanced reasoning and tool use

Strong long-context and reasoning capabilities

High-quality reasoning and instruction following

Thinking/reasoning variant of Qwen3 235B

Compact model balancing quality and speed

Efficient MoE model for instruction following

Google's Gemma 3 instruction-tuned model

Powerful instruction-following model with long-context capabilities

Highly capable model fine-tuned for multi-turn conversations

Third-generation model trained via decentralized compute

Massive MoE model for demanding reasoning and agentic workloads

Hybrid MoE model optimized for efficient multi-agent AI

Open, efficient omni-modal reasoning model for agentic AI

Super reasoning model for complex multi-step tasks

Vision-language model supporting text and images

Efficient vision-language model with strong multimodal capabilities

High-precision dense retrieval with multilingual coverage (4,096 dims)