Key Takeaways:

• SambaNova has added function calling support to its DeepSeek-R1-0528 model, making agent workflows more dynamic on SambaCloud.
• DeepSeek-R1-0528 combines strong reasoning capabilities with transparent access to reasoning tokens, setting it apart from proprietary models.
• Function calling allows AI agents to trigger custom tools, fetch live data, and integrate results directly into conversations.
• Developers can implement function calling on SambaCloud in five steps: log in, define the schema, configure requests, handle tool calls, and return results.
• CrewAI, an open-source multi-agent framework, can work with DeepSeek-R1-0528 to orchestrate team-based agent workflows.

SambaNova has introduced function calling to its DeepSeek-R1-0528 checkpoint, opening the door for more powerful AI agent workflows on SambaCloud. This update brings a major capability often associated with proprietary models like OpenAI’s GPT-4 or Google’s Gemini, but with a twist: DeepSeek remains open and transparent about its reasoning tokens, giving developers an extra layer of insight into model behavior.

Why this update matters

DeepSeek-R1 first gained attention when it was released in January, offering reasoning capabilities that began to rival closed-source systems. The latest checkpoint, DeepSeek-R1-0528, builds on that by enabling function calling. This allows developers to design applications where an AI doesn’t just answer questions but can take real actions—retrieving external data, updating records, or running processes—before returning results as part of the same conversation.

The blog notes that “function calling lets your AI not only respond to prompts but also trigger custom functions to fetch data, perform actions, and seamlessly integrate the outputs back into the conversation.” This distinction is critical. Instead of relying solely on static training data or approximations, AI agents can now connect directly to live systems, extending their utility far beyond text generation.

How function calling works on SambaCloud

The implementation process follows five steps. First, developers log into SambaCloud with their API credentials. Next, they define the function schema, usually in JSON format, outlining parameters such as function name, description, and required inputs. The third step involves configuring a request that includes this schema so the model knows what tools it can call. Once a function is invoked, developers must capture the tool call output, run the actual function in their system, and finally return the results to the model so it can respond with the updated information.

To make the concept tangible, the blog provides a sample weather lookup tool. While the example uses randomized data instead of a live API, the flow demonstrates exactly how an agent could call the function, receive a structured response, and continue the dialogue using the new information.

Open source transparency meets enterprise utility

A defining feature of DeepSeek-R1 is its use of “reasoning tokens,” which allow developers to inspect the model’s thought process. Proprietary systems rarely expose this level of visibility, meaning teams often operate without clarity on why a model responded the way it did. SambaNova positions this as a key differentiator. By marrying reasoning transparency with practical features like function calling, the company hopes to give enterprises both confidence and flexibility as they deploy agentic AI.

The blog emphasizes that DeepSeek-R1-0528 “matches the reasoning capability of leading proprietary models while remaining fully open-sourced and checkpointed.” For organizations seeking both performance and openness, this combination could prove compelling.

Pairing with CrewAI for multi-agent workflows

While function calling enables a single model to interact with external systems, the next leap in capability comes from orchestrating multiple agents. CrewAI, an open-source framework highlighted in the post, provides this orchestration layer. Developers can set up roles—such as a researcher, planner, and writer—that work together, each with access to different tools through function calls. DeepSeek-R1-0528 integrates into this setup, enabling more complex workflows that resemble collaborative human teams.

In practice, this means an enterprise could design an agent crew where one agent gathers market data, another drafts an analysis, and a third refines the output—all powered by function calls that connect to live systems. The blog notes that “CrewAI allows you to define teams of agents that collaborate, delegate tasks, and interact with tools seamlessly,” making DeepSeek’s function calling a natural fit.

Developer accessibility

Getting started is straightforward for teams already familiar with API workflows. The blog provides code snippets showing how to initialize the client:

client = openai.OpenAI(
base_url=”https://api.sambanova.ai/v1″,
api_key=”YOUR_SAMBANOVA_API_KEY”
)

From there, developers pass their function schema into a chat.completions.create call, allowing the model to issue structured tool requests. Capturing the function’s arguments and returning the result completes the loop, transforming static chat into active, tool-integrated reasoning.

The broader context

SambaNova’s addition of function calling reflects a broader trend in AI: moving from models as knowledge bases to models as agents. Function calling is becoming a baseline expectation for enterprise-grade systems. What makes this update notable is that SambaNova delivers it through an open model with transparent reasoning. That approach could resonate with companies seeking to balance cutting-edge performance with accountability.

Conclusion

By adding function calling to DeepSeek-R1-0528, SambaNova has strengthened the agentic capabilities of its platform while maintaining its commitment to openness. Developers can now create AI systems that both reason and act, integrating seamlessly with business workflows. Paired with frameworks like CrewAI, these tools make it possible to design collaborative, multi-step AI processes. As enterprises explore how to operationalize AI safely and effectively, this update offers a flexible and transparent option for moving from static chatbots to dynamic, agentic systems.

Learn how AI Agents can supercharge your company’s profits and productivity at TMC’s AI Agent Event Sept 29-30, 2025 in DC.

Rich Tehrani serves as CEO of TMC and chairman of ITEXPO #TECHSUPERSHOW Feb 10-12, 2026 and is CEO of RT Advisors and is a Registered Representative (investment banker) with and offering securities through Four Points Capital Partners LLC (Four Points) (Member FINRA/SIPC). He handles capital/debt raises as well as M&A. RT Advisors is not owned by Four Points.

The above is not an endorsement or recommendation to buy/sell any security or sector mentioned. No companies mentioned above are current or past clients of RT Advisors.

The views and opinions expressed above are those of the participants. While believed to be reliable, the information has not been independently verified for accuracy. Any broad, general statements made herein are provided for context only and should not be construed as exhaustive or universally applicable.

Portions of this article may have been developed with the assistance of artificial intelligence, which may have contributed to ideation, content generation, factual review, or editing.