Payment ops
For partner-bank RFIs, our top pick is Kimi K3 from Moonshot AI. Native text, image, and video input with a 1M-token context, so it reads the whole RFI packet in one pass. GLM-5.3 Flash is a close second, and MiniMax M3 rounds out the list.
Partner banks send requests for information, hold-harmless letters, and return-of-funds cases as emails and PDFs. An agent has to read the documents, pull the payment and KYC records, and draft a reply, so document and image understanding matter as much as reasoning.
Updated
Native text, image, and video input with a 1M-token context, so it reads the whole RFI packet in one pass.
The first open 3T-class model, natively multimodal with a 1M-token context.
Input: text, image, video
Harnesses in Runtime
Image input at 18B active parameters and an MIT license, a cost-effective default for high RFI volume.
A natively multimodal GLM that Z.ai says beats GLM-5.2 at one-tenth the price.
Input: text, image
Harnesses in Runtime
Multimodal with sparse attention built for million-token contexts, useful for long correspondence threads.
A natively multimodal model with sparse attention built for million-token contexts.
Input: text, image, video
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 partner-bank RFIs, our top pick is Kimi K3 from Moonshot AI. Native text, image, and video input with a 1M-token context, so it reads the whole RFI packet in one pass. GLM-5.3 Flash is a close second, and MiniMax M3 rounds out the list.
Reads PDFs, scans, and screenshots. Long context for full case files. Reliable drafting for an approval step. 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.
GLM-5.3 Flash (MIT) is permissive and fine for commercial use. Kimi K3 (Kimi K3 License), MiniMax M3 (MiniMax Community License) use custom licenses, so read the terms first.
