Run OpenCode Offline: Ollama Setup, Context Windows & Custom Models
This walkthrough shows how to run OpenCode fully offline by pairing it with Ollama and the Qwen 3.6 local model, covering install via Homebrew or npm, tuning Ollama's context window for load speed versus capacity, saving custom models with a pinned context size, and registering the model in OpenCode's JSON config so it appears automatically.
Darren Builds AI10 minTranscript found
Quick learning frame
Read this before watching.
A local runtime lesson is about fit: model, quantization, hardware, endpoint, latency, privacy, tool integration, and task limits.
New playlist item from Darren Builds AI; queued for transcript-backed review, topic mapping, and a practical learning artifact.
Skill you build: The ability to set up and tune a local open source coding model (Ollama plus OpenCode), matching context window size to the task and machine, and wiring custom models into OpenCode's provider config.
Watch for the shift from claim to mechanism. The learning value is the point where the transcript reveals a repeatable action, tool boundary, context move, review habit, or artifact.
Concept diagram
Where this video fits.
01Task
02Hardware
03Model/quantization
04Runtime endpoint
05Agent tool loop
06Benchmark task
07Fallback
Deep lesson
Turn this video into working knowledge.
2,031 cleaned transcript words reviewed across 586 timed caption segments.
Thesis
Run OpenCode Offline: Ollama Setup, Context Windows & Custom Models teaches a practical local model/runtime move: This walkthrough shows how to run OpenCode fully offline by pairing it with Ollama and the Qwen 3.6 local model, covering install via Homebrew or npm, tuning Ollama's context window for load speed versus capacity, saving custom models with a pinned context size, and registering the model in OpenCode's JSON config so it appears automatically.
The goal is not to remember the video. The goal is to extract the operating principle, tie it to timestamped evidence, test how far the claim transfers, and make something reusable.
0:00
Ollama plus OpenCode setup
“Do you want to use an open source model on your local machine to work with open code where you can actually reliably code and actually perform tasks? And are you also looking for a few tips that...”
You need two installs, Ollama (direct download) and OpenCode (npm, Bun, Yarn, or Homebrew), then pull Qwen 3.6 with 'ollama run': it is a 24 GB model that runs fine on 64 GB unified memory and workable on 32 GB, while smaller options like the 27B variant or Qwen 3.5 9B tested unreliable, so stick with 3.6 for results. Install both tools, run 'ollama run qwen3.6' to pull the model, and verify both 'ollama' and 'opencode' launch from your terminal before doing anything else.
3:49
Context window tradeoffs
“uh uh, already there, which is great cuz now I can actually start using this. For example, it'll One thing to note, this model can take a while to load up with your open code session. All right,...”
Ollama's settings menu now lets you change the context window, but bigger windows (128K or 256K) make the model load much slower and can break machines that cannot handle them; Darren uses 32K for quick function work, had to bump to 64K when the default build agent ran out, and notes Qwen 3.6's full context is 256K per its Ollama model page. Pick one coding task, set the context window to the smallest size that fits it (for example 32K or 64K), run the task, and note when you actually hit the threshold before increasing.
7:38
Custom models and config
“with Darren's model latest. And this will always have this context set to 65K, which is really great. So, I if I want to upload this instead of config file for open code, I can do that. One...”
You can pin a context size permanently by running the model, using set parameter (for example 65K), and saving it under a new name that then appears in 'ollama list'; to make any Ollama model show up inside OpenCode without the 'ollama launch opencode' path, add an Ollama provider entry using the npm AI SDK OpenAI-compatible package and the model name to your .opencode.json config. Create your own named model with a pinned context size via set parameter and save, then add it to .opencode.json so it appears in OpenCode's model list on a plain launch.
01
Task
Start with this video's job: This walkthrough shows how to run OpenCode fully offline by pairing it with Ollama and the Qwen 3.6 local model, covering install via Homebrew or npm, tuning Ollama's context window for load speed versus capacity, saving custom models with a pinned context size, and registering the model in OpenCode's JSON config so it appears automatically. Treat "Task" as the outcome you are trying to make visible, not a topic label. Anchor it to 0:00, where the video says: “Do you want to use an open source model on your local machine to work with open code where you can actually reliably code and actually perform tasks? And are you also looking for a few tips that...”
02
Hardware
Use "Hardware" to locate the part of the local model/runtime mechanism the video is demonstrating. Ask what changes in your real setup if this claim is true. Anchor it to 3:49, where the video says: “uh uh, already there, which is great cuz now I can actually start using this. For example, it'll One thing to note, this model can take a while to load up with your open code session. All right,...”
03
Model/quantization
Turn "Model/quantization" into the reusable artifact for this lesson: A local model/runtime fit sheet with hardware constraints, model choice, endpoint setup, task benchmark, and fallback rule. This is where watching becomes something you can inspect and reuse.
04
Runtime endpoint
Use "Runtime endpoint" as the application surface. Decide whether the idea touches a browser flow, a local file, a model choice, a source document, a UI, or a review step.
05
Agent tool loop
Use "Agent tool loop" to prove the lesson. The evidence should connect back to the video title, transcript anchors, and a concrete output, not a generic best-practice claim.
06
Benchmark task
Use "Benchmark task" to carry the idea forward: save the prompt, checklist, diagram, or operating rule that would make the next agent run better.
07
Fallback
Connect "Fallback" to Run OpenCode Offline: Ollama Setup, Context Windows & Custom Models by naming the claim, the evidence, and the artifact it should produce.
Example
Source-backed artifact packet
Convert the video into a scoped artifact request that includes the transcript claim, mechanism, acceptance criteria, and proof. The output should be a local model/runtime fit sheet with hardware constraints, model choice, endpoint setup, task benchmark, and fallback rule..
Example
Local model/runtime proof brief
Separate what the speaker claims, what the demo actually proves, and what still needs outside verification before you adopt the local model/runtime pattern.
Example
Teach-back module
Transform the lesson into a definition, a Task -> Hardware -> Model/quantization -> Runtime endpoint -> Agent tool loop -> Benchmark task -> Fallback diagram, one misconception, one practice exercise, and a check-for-understanding question.
Do not learn it wrong
Treating the title as the lesson without checking what the transcript actually says.
using a local model like ChatGPT
ignoring latency/context limits
no benchmark task
Letting the lesson drift into local-model ideology.
Letting the lesson drift into hardware specs without workflow fit.
Letting the lesson drift into benchmarks unrelated to the actual task.
Do not count this as learned until these are true.
01
State the transcript-backed claim in your own words: This walkthrough shows how to run OpenCode fully offline by pairing it with Ollama and the Qwen 3.6 local model, covering install via Homebrew or npm, tuning Ollama's context window for load speed versus capacity, saving custom models with a pinned context size, and registering the model in OpenCode's JSON config so it appears automatically.
02
Explain the practical stakes without hype: New playlist item from Darren Builds AI; queued for transcript-backed review, topic mapping, and a practical learning artifact.
03
Map the idea onto the Task -> Hardware -> Model/quantization -> Runtime endpoint -> Agent tool loop -> Benchmark task -> Fallback sequence and name the weakest link.
04
Produce the artifact and include the evidence that proves it: A local model/runtime fit sheet with hardware constraints, model choice, endpoint setup, task benchmark, and fallback rule.
Put it into practice
Give this grounded prompt to Codex or Claude after watching.
You are helping me turn one specific YouTube video into real, durable learning.
Source video:
- Title: Run OpenCode Offline: Ollama Setup, Context Windows & Custom Models
- URL: https://www.youtube.com/watch?v=uBmF5I_4gWg
- Topic: Codex + Claude Workflows
- My current learning frame: Set up Ollama with Qwen 3.6 and OpenCode on your machine, build a small Next.js landing page at a 32K context window, then bump to 64K and register a custom-named model in .opencode.json, comparing load time and how much guidance the local model needs versus a hosted one.
- Why this matters: New playlist item from Darren Builds AI; queued for transcript-backed review, topic mapping, and a practical learning artifact.
Transcript anchors from this exact video:
- 0:00 / Evidence 1: "Do you want to use an open source model on your local machine to work with open code where you can actually reliably code and actually perform tasks? And are you also looking for a few tips that..."
- 1:32 / Evidence 2: "going to be looking at running a local model from Ollama. And what we're going to be doing is we're actually going to be using the Qwen 3.6 model. So, if you just go Qwen 3.6 in Ollama..."
- 3:49 / Evidence 3: "uh uh, already there, which is great cuz now I can actually start using this. For example, it'll One thing to note, this model can take a while to load up with your open code session. All right,..."
- 5:38 / Evidence 4: "can always increase the performance and then, sorry, the context window, and it'll just take a bit of time to load up, but at least you can use it that way. This is the site that the local..."
- 7:38 / Evidence 5: "with Darren's model latest. And this will always have this context set to 65K, which is really great. So, I if I want to upload this instead of config file for open code, I can do that. One..."
- 9:39 / Evidence 6: "local models. Also, you're going to probably have to have some custom agents really trim it down and have uh system prompts that actually make it a lot better. Anyway, I hope you guys really enjoyed this video..."
Video-aware target:
- Prompt lane: Local model/runtime
- Mechanism to extract: Identify why the local setup works or fails for this specific agent task, not whether local models are generally good.
- Artifact to produce: A local model/runtime fit sheet with hardware constraints, model choice, endpoint setup, task benchmark, and fallback rule.
- Artifact must include: hardware; runtime; model/quantization; endpoint; agent integration; benchmark/fallback
Your task:
1. Use the transcript anchors above as the primary source packet. If you add outside context, label it clearly as outside context and keep it secondary.
2. Create a source-check table with columns: timestamp, claim, transcript support, what the demo proves, confidence, and what still needs verification.
3. Extract the actual teachable mechanism from the video: Identify why the local setup works or fails for this specific agent task, not whether local models are generally good. Do not invent claims that are not supported by the title, lesson frame, or transcript anchors.
4. Build a reusable learning artifact: A local model/runtime fit sheet with hardware constraints, model choice, endpoint setup, task benchmark, and fallback rule.
5. Include:
- a plain-English definition of the core idea
- a diagram or structured model using this sequence: Task -> Hardware -> Model/quantization -> Runtime endpoint -> Agent tool loop -> Benchmark task -> Fallback
- answers to these source questions: What machine/runtime is shown? | What task exposes the model limit? | What setup change improves the loop?
- 3 concrete examples that apply the video idea to real agentic work, such as Ollama or LM Studio coding endpoint; MLX Apple Silicon runner; DGX-backed Hermes session
- 2 failure modes the video helps prevent, chosen from the transcript evidence and these likely risks: using a local model like ChatGPT; ignoring latency/context limits; no benchmark task
- a checklist for the next real workflow, focused on: task fit, runtime setup, latency/context, tool loop, fallback
- one practical exercise with a clear done signal: Choose one real coding task and specify the pass/fail benchmark for a local model.
6. Add a "learning transfer" section: what changes in my workflow tomorrow if I actually learned this?
7. Add a "source check" section that cites which transcript anchor supports each major takeaway.
Quality bar:
- Make this specific to "Run OpenCode Offline: Ollama Setup, Context Windows & Custom Models", not a generic Codex + Claude Workflows essay.
- Tie each harness element to a transcript anchor that names a tool, state boundary, permission, model behavior, or verification step.
- Prefer operational examples, failure modes, and reusable artifacts over broad definitions.
- Call out uncertainty instead of smoothing over weak evidence.
- Avoid these generic drifts: local-model ideology; hardware specs without workflow fit; benchmarks unrelated to the actual task.
- If evidence is weak or missing, stop and say what transcript segment or timestamp needs review instead of guessing.
- Finish with a concise artifact I could paste into my learning app.
Misconceptions
What to stop believing.
One agent should do every task.
Different tools have different strengths. Routing is part of the workflow.
More context is always better.
Relevant context helps; stale context causes drift and cost.
Practice studio
Learning only counts when you make something.
01
Transcript evidence map
Separate what the video actually says from what you already believe about the topic.
3 source-backed takeaways with timestamps, confidence, and a transfer note.02
One useful artifact
Apply the video to a real workflow and produce a local model/runtime fit sheet with hardware constraints, model choice, endpoint setup, task benchmark, and fallback rule..
A reusable artifact with a done signal and one verification step.03
Local model/runtime teach-back card
Explain the local model/runtime mechanism to someone who has not watched the video yet.
A 90-second explanation, one diagram, one example, and one misconception to avoid.
Recall check
Answer first, then reveal — without rewatching.
Which local model does the video recommend for reliable coding with OpenCode, and why not smaller ones?
What tradeoff comes with increasing Ollama's context window setting?
Why might a Qwen model not appear in OpenCode's model list, and how do you fix it?
Source shelf
Use the video as a doorway, then verify with primary sources.
