Reads text, images, and video with a 1M-token context, so a whole dispute file fits in one run.
A natively multimodal model with sparse attention built for million-token contexts.
Input: text, image, video
Harnesses in Runtime
Risk and fraud
For chargebacks and disputes, our top pick is MiniMax M3 from MiniMax. Reads text, images, and video with a 1M-token context, so a whole dispute file fits in one run. Kimi K3 is a close second, and GLM-5.3 rounds out the list.
Dispute agents gather delivery proof, order history, customer messages, and screenshots, then assemble an evidence packet before the network deadline. That means mixed inputs, long context, and careful writing.
Updated
Reads text, images, and video with a 1M-token context, so a whole dispute file fits in one run.
A natively multimodal model with sparse attention built for million-token contexts.
Input: text, image, video
Harnesses in Runtime
The most capable multimodal open model for complex, document-heavy cases.
The first open 3T-class model, natively multimodal with a 1M-token context.
Input: text, image, video
Harnesses in Runtime
Strong at long, multi-tool runs that pull from the processor, shipping data, and the help desk.
Z.ai's strongest open-weight model, post-trained for complex coding and long-horizon agent work.
Input: text
Harnesses in Runtime
Turn the steps your team already follows into a skill the agent reads before every run.
Run two or three of these models against past cases. Evals show which one passes at the lowest cost.
The agent works on its own computer in your VPC, with role-based access and an approval before sensitive actions.
For chargebacks and disputes, our top pick is MiniMax M3 from MiniMax. Reads text, images, and video with a 1M-token context, so a whole dispute file fits in one run. Kimi K3 is a close second, and GLM-5.3 rounds out the list.
Image input for receipts and screenshots. Long context for full order histories. Clear, structured writing for evidence packets. Test two or three candidates against your own past cases before you commit, and keep a human approval on any step that moves money or changes a customer's status.
Yes. Download the weights and serve them with vLLM or SGLang in your own cloud, or use Amazon Bedrock or Google Vertex AI in your account. With Runtime, the agent's computer also runs in your VPC, so card data and PII stay in your environment.
MiniMax M3 (MiniMax Community License), Kimi K3 (Kimi K3 License), GLM-5.3 (GLM-5.3 License) use custom licenses, so read the terms first.