Weights & Biases × WorkDaddy: LLM Provider Integration (2026)
Use Weights & Biases as the engine behind your WorkDaddy agent team: 31 models available, context windows up to 1M tokens, connected in minutes with your own key.
Weights & Biases is one of 178 LLM providers supported by WorkDaddy’s model layer in 2026, in the Enterprise Data Platforms group. The catalog currently lists 31 models from Weights & Biases: 20 with explicit reasoning, 10 with vision input, and 31 with tool calling — the capability that matters most for agent work. The largest context window reaches 1M tokens.
At a glance
| Models available | 31 |
|---|---|
| Reasoning models | 20 |
| Vision models | 10 |
| Tool-calling models | 31 |
| Largest context window | 1M tokens |
| API endpoint | https://api.inference.wandb.ai/v1 |
| API key variable | WANDB_API_KEY |
| SDK adapter | @ai-sdk/openai-compatible |
| Provider docs | docs.wandb.ai ↗ |
Why run WorkDaddy agents on Weights & Biases in 2026
Tool calling, context size, and cost decide how well a provider drives an agent team. Weights & Biases brings 31 tool-calling models to WorkDaddy — enough to power the full loop of storefront edits, analytics queries, and content production — with 1M tokens of context at the top end for whole-codebase and long-report work. Because WorkDaddy is model-agnostic, you can route heavy reasoning to Weights & Biases’s strongest model and bulk work to its cheapest, inside one team.
Best Weights & Biases models for e-commerce agents (2026)
The current flagships from Weights & Biases in the WorkDaddy catalog, newest first:
Kimi K3 — Kimi K3 is a 2.8T-parameter multimodal MoE model with 104B active parameters built for long-horizon coding and agentic workflows. (1M context)
GLM 5.2 — GLM-5.2 is a Mixture-of-Experts language model featuring 40 billion activated parameters and a total of 744 billion parameters. (262K context)
MiniMax M3 — MiniMax M3 is a multimodal MoE model with 23B active parameters optimized for coding and agentic workflows. (262K context)
Kimi K2.7 Code — Kimi K2.7 Code is a 1T-parameter MoE model with 32B active parameters purpose-built for long-horizon agentic coding and software engineering. (262K context)
Nemotron 3 Ultra — Nemotron 3 Ultra is a powerful MoE model designed for long-running agents across coding, deep research, and enterprise automation. (262K context)
Mellum2 12B A2.5B — Mellum2-12B-A2.5B-Instruct is a fast MoE model with 131K context built for coding, tool use, and low-latency AI workflows. (131K context)
How to connect Weights & Biases to WorkDaddy
WorkDaddy talks to Weights & Biases at api.inference.wandb.ai via the @ai-sdk/openai-compatible adapter. Add your credential (typically WANDB_API_KEY) in Settings → Models, pick a default model, and every agent can use it immediately — or add Weights & Biases as one engine among several and let tasks route to the best fit.
Featured models
| Model | Context | Reasoning | Released |
|---|---|---|---|
| Kimi K3 | 1M | Yes | 2026-07-16 |
| GLM 5.2 | 262K | Yes | 2026-06-16 |
| MiniMax M3 | 262K | Yes | 2026-06-12 |
| Kimi K2.7 Code | 262K | Yes | 2026-06-12 |
| Nemotron 3 Ultra | 262K | Yes | 2026-06-04 |
| Mellum2 12B A2.5B | 131K | No | 2026-06-01 |
How it works
Connect
In WorkDaddy, open Settings → Models, choose Weights & Biases, and paste your API key (WANDB_API_KEY). No key? Start on the managed gateway instead.
Put agents to work
Pick Kimi K3 or any of the 31 available models as your default — per-agent overrides let you match model to task.
Review and approve
Agent output stays draft-first regardless of the model: review diffs and approve actions exactly as before.
Frequently asked questions
Does WorkDaddy support Weights & Biases in 2026?
Yes — Weights & Biases is a supported LLM provider with 31 models in the WorkDaddy catalog, connected with your own API key via the @ai-sdk/openai-compatible adapter.
How many Weights & Biases models can I use with WorkDaddy?
31 models are listed for Weights & Biases, including 20 reasoning models and 10 vision-capable models. The flagship lineup above shows the newest.
What context window do Weights & Biases models offer?
Up to 1M tokens on the largest model — relevant for whole-repo storefront work and long analytics reports.
Is Weights & Biases the best LLM provider for e-commerce agents?
It depends on the task mix — Weights & Biases sits in the Enterprise Data Platforms group. WorkDaddy is model-agnostic, so the practical answer in 2026 is to combine providers: route each agent task to whichever model fits best.
Keep reading
SAP AI Core
Enterprise Data Platforms
Neon
Enterprise Data Platforms
Databricks
Enterprise Data Platforms
Snowflake Cortex
Enterprise Data Platforms
Explore more integration directories
Put the team to work on your store
Connect your storefront, analytics, and email stack, set a goal, and let the agents run the work end to end. Start on the free plan with your own model key.