Ollama now supports subagents and web search in Claude Code! Subagents can run tasks in parallel, such as file search, code exploration, and research, each in their own context. No MCP servers to configure or API keys required. Try it with any model on Ollama's cloud: ollama launch claude... --model minimax-m2.5:cloudshow more

Longer coding sessions stay productive. Side tasks don’t fill the context with noise. Some models will naturally trigger subagents when needed (minimax-m2.5, glm-5, kimi-k2.5), but you can force triggering subagents by telling the model to “use/spawn/create subagents” Example prompts: > spawn subagents to explore the auth flow, payment integration, and notification system > audit security issues, find performance bottlenecks, and check accessibility in parallel with subagents > create subagents to map the database queries, trace the API routes, and catalog error handling patterns Blog post:

Ollama’s web search is now built into the Anthropic compatibility layer. When a model needs current information, Ollama handles the search and returns results directly without any additional configuration. Subagents can leverage web search to research topics in parallel and come back with actionable results. > research the postgres 18 release notes, audit our queries for deprecated patterns, and create migration tasks > create 3 research agents to research how our top 3 competitors price their API tiers, compare against our current pricing, and draft recommendations > study how top open source projects handle their release process, review our CI/CD pipeline, and draft improvements Blog post:

daily driver confirmed ✅

My new ride!

10/10. how much?

I just put in $20 for a month to try it. Looking to upgrade to Ollama’s $100 plan.

🧧🧧🧧 give me some and then we’ll talk

The parallel context isolation is the real game changer here. Been running into context pollution issues with complex tasks - having subagents each maintain their own context means the main agent stays focused. No MCP config = huge win for onboarding new devs to AI-assisted workflows too.

Local model integration with developer tools enabling parallel task execution and web research capabilities expands AI coding assistant functionality without external API dependencies.

that's very cool!

Ollama subagents + Claude Code = department-level parallel coding army 🔥 One prompt → codebase exploration + security audit + perf bottlenecks run in parallel, isolated contexts, zero pollution. Zero friction: `ollama launch claude --model minimax-m2.5:cloud` → instant, built-in web search, frontier perf at fraction of cost. Manual grunt work is dead. Strategy mode ON. Killer update @ollama

This is huge for local development! Subagents running in parallel without MCP setup removes so much friction. how does the subagent context isolation work? Are they fully independent model calls or do they share some base context? And does minimax-m2.5:cloud route to their API or is it actually running locally?

I want to cover it on my channel. can I get some credits to ollama cloud? thanks

Big upgrade for the ecosystem 🚀 Parallel subagents + built-in web search — all without MCP setup or API keys — is a serious unlock for developers. Excited to see what teams build on Ollama Cloud with this. Let’s explore collaboration opportunities.

Huge update 👏 Subagents + web search inside Claude Code on Ollama is a serious productivity unlock.

thank you

Coin it @clanker_world Name: Ollama Clanker Ticker: Ollama

I just used that command ollama launch claude --model minimax-m2.5:cloud to start a coding project. Minimax is so good

@OsayKancuno How that works?

this is so amazing

Subagents running in parallel contexts is the interesting shift here. We’re moving from single-threaded chat interfaces to orchestration layers. Once you combine parallel task execution + web search, the model becomes more of a coordination engine than just a responder.

Not responding well, codex better

subagents running in parallel is a game changer. been using claude code for a few weeks now and the context management alone saves so much time. no MCP config needed is the cherry on top

@grok summary ollama with this

@itsaibot #detect

it's way slower than I expected

the zero config part is huge. but curious how local models handle the subagent reasoning quality, since subagents need to make autonomous decisions without human review. that's where smaller models tend to fall apart.

Everyone should read this and support Ollama, open source is now more important than ever.

Impressive progress! For local dev, how do you handle resource constraints when running multiple subagents in parallel? Is there built-in throttling or does it rely on the host's available RAM/GPU? Curious how this scales on laptops vs desktops.

Hey @ollama , could you support an embedding model on your cloud models?

My laptop is about to reach the temperature of the sun just so I do not have to copy and paste an API key. Worth it.

no MCP config, no API keys, subagents just work out of the box?? this is the kind of DX that makes you never go back. been waiting for parallel task execution in Claude Code forever

Subagents aren't just parallel search — each gets its own context, keeping main sessions clean across multi-hour tasks. That's the real unlock for complex workflows. The bigger picture: Claude Code's workflow layer (skills, subagents, MCP) is becoming model-agnostic infrastructure. We run 11 skill files daily. The orchestration outlasts any specific model.

subagents running in parallel without MCP config is a massive quality of life improvement. no more wrestling with server setup just to get basic tool use working

The parallel execution is huge. I've been patching together multiple API calls for my SaaS automation workflows - this could eliminate so much complexity. No MCP configuration is exactly what solo devs need when every hour counts.

While parallel tasks are a game changer, I'm still waiting to see how subagents handle conflicting priorities. Doesn't the simplicity of no API keys required come with potential tradeoffs in customization? Simplifying the process is a step forward, but at what cost to advanced users.

Frictionless setup drives adoption.