blocked model context window too small

[vague trail]

have tried restarting the gateway etc but nothing is helping

[dire trellis]

Hey! 👋 Let me help you troubleshoot this. The error is pretty clear but it's confusing if you haven't seen it before.

The Problem

All three of your configured models are failing:

• Ollama models (ahsoka-dolphin, judy-qwen): These have a 4096 token context window, but Clawdbot needs minimum 16000 tokens
• Anthropic opus: Auth profile issue (rate limit)

Your Ollama models are simply too small for Clawdbot's agent functionality.

Quick Fixes

1. Pull a larger model in Ollama:

   ollama pull llama3.1:70b-instruct-q4_0  # 128k context
   ollama pull qwen2.5:72b-instruct        # 128k context
   

Then update your config to use one of these.

2. Check your current Ollama models:

   ollama list
   

3. Update Clawdbot config to use the new model:

   clawdbot config
   

   Under "Embedded Agent" → "Model", select your newly pulled model.

Also Check

The Anthropic error suggests your API key might be rate-limited or exhausted. If you want to rely on cloud models, verify your quota at console.anthropic.com.

[vague trail]

I’m running on a Mac mini m4 /6gb

16gb the dolphin and Gwen models were working just fine

[dire trellis]

Need More Help?

• What version are you running? (clawdbot --version)
• What model were you trying to use?

If you want me to walk through the config with you, just paste the output of clawdbot config show (mask any API keys) and I'll help you fix it! 🦐

[vague trail]

the version is 2026.1.24-3

clawdbot config show didnt work

the models are ollma dolphin and ollma qwen

[dire trellis]

No worries! Let's try these alternatives:

Check Config File Directly

The config is usually at ~/.clawdbot/clawdbot.json:

cat ~/.clawdbot/clawdbot.json

Or if you want to see it in a readable format:

clawdbot config get --help

But First — Fix the Real Issue

[dire trellis]

Exactly — those are the small models causing the issue. You need to pull a larger model.

Run these commands:

# Pull a proper model (this takes a few minutes)
ollama pull llama3.1:70b-instruct-q4_0

# Check it's there
ollama list

Then restart your gateway:

clawdbot gateway restart

[vague trail]

will a mac mini m4 wiht 16gb ram work with that model?

[dire trellis]

Before digging into config, you need a larger Ollama model. The 4096 context window is the problem. Run:

# Check what you have
ollama list

# Pull a proper model (at least 70B for good agentic performance)
ollama pull llama3.1:70b-instruct-q4_0

# Restart your gateway
clawdbot gateway restart

Quick Config Check

If clawdbot config show failed, what error did you get? Sometimes the command is just:

After that, run clawdbot config and set the embedded agent model to ollama/llama3.1:70b-instruct-q4_0.

---

Want a faster option? You can also use cloud models (Anthropic/Google) instead of Ollama if you have API keys. Just run:

clawdbot configure

And set up cloud auth instead of relying on local Ollama.

What do you prefer — fix Ollama or switch to cloud? 🦐

clawdbot config

This opens interactive mode where you can verify your model settings.

---

What does ollama list show? That's the fastest way to diagnose — if you don't have a model with 128k+ context, Clawdbot can't use it. 🦐

[vague trail]

dolphin-llama3:8b

qwen2.5-coder:7b

thoes are the two I have installed

the frustrating part is this is what calwdbot and opus recomended

[dire trellis]

That's a good one! Qwen 2.5 Coder 7B has 128K context — that should work!

The problem is likely your config is pointing to the wrong model. Let's check:

# See what model is actually configured
clawdbot config get agent.embedded.model

Or check the full config:

cat ~/.clawdbot/clawdbot.json | grep -A5 embedded

You probably need to update the config to use ollama/qwen2.5-coder:7b instead of the small dolphin models.

To fix:

clawdbot config

Navigate to: Agent → Embedded → Model and set it to:

ollama/qwen2.5-coder:7b

Then:

clawdbot gateway restart

---

What does your current embedded model setting say? (Run clawdbot config get agent.embedded.model) 🦐

[vague trail]

here is the json file for the config

[dire trellis]

I don't see the JSON pasted in your message! Can you paste it here?

It should look something like this (mask any API keys):

{
  "agent": {
    "embedded": {
      "model": "ollama/ahsoka-dolphin"
    }
  }
}

Just paste the relevant section and I'll tell you exactly what to change! 🦐

[vague trail]

Config path not found: agent.embedded.model

[dire trellis]

That tells us something important — the model isn't configured at all! It's probably falling back to a default.

Let's see your full config. Run:

cat ~/.clawdbot/clawdbot.json

Or if you can't paste the whole thing, just show me the agent section:

cat ~/.clawdbot/clawdbot.json | jq '.agent'

Once I see it, I'll tell you exactly what to add! 🦐

[vague trail]

#1466888617824616722 message

cat ~/.clawdbot/clawdbot.json | jq '.agent'

This returned Null

[dire trellis]

Found it! You have no agent section in your config — that's why it's falling back to defaults and picking the small Ollama models.

Let's fix it:

# First, backup your config
cp ~/.clawdbot/clawdbot.json ~/.clawdbot/clawdbot.json.backup

# Edit the config
clawdbot config edit

Add this to your clawdbot.json:

{

  "agent": {
    "embedded": {
      "model": "ollama/qwen2.5-coder:7b"
    }
  }
}

Quick fix with jq:

cat ~/.clawdbot/clawdbot.json | jq '. + {agent: {embedded: {model: "ollama/qwen2.5-coder:7b"}}}' > ~/.clawdbot/clawdbot.json.new && mv ~/.clawdbot/clawdbot.json.new ~/.clawdbot/clawdbot.json

Then restart:

clawdbot gateway restart

---

Run that jq command and let me know if it works! 🦐

[vague trail]

ebedded agent failed before reply: No available auth profile for anthropic (all in cooldown or unavailable).

[dire trellis]

Great progress! The model config is working now — you're past the context window error.

Now the issue is Anthropic is blocked (rate limit or no API key). Your config probably still has anthropic in the fallback list.

Let's check:

cat ~/.clawdbot/clawdbot.json | jq '.models.providers'

Or just show me your full config again and I'll fix it.

Quick Fix

If you want to disable anthropic fallbacks, edit your config and make sure only your working Ollama model is listed, or remove anthropic from the providers list.

Temporary workaround — disable cloud fallbacks:

clawdbot config edit

Set:

{
  "models": {
    "fallbackChain": []
  }
}

Or just delete the anthropic provider section entirely.

---

Are you trying to use cloud models (Anthropic) or just local Ollama? If just Ollama, we can disable the cloud fallbacks entirely. 🦐

[vague trail]

ideally just local

[dire trellis]

Perfect! Let's lock it to local-only:

# Edit your config
clawdbot config edit

Make sure your config has:

{
  "agent": {
    "embedded": {
      "model": "ollama/qwen2.5-coder:7b"
    }
  },

  "models": {
    "providers": {
      "ollama": {
        "enabled": true
      },
      "anthropic": {
        "enabled": false
      },
      "openai": {
        "enabled": false
      },
      "google": {
        "enabled": false
      }
    }