Context size exceeded
Error
request (10669 tokens) exceeds the available context size (8192 tokens)Why this happens
The conversation history plus your current message is larger than the model's context window. Smaller models (8k context) hit this quickly with multi-file conversations.
Fix
1. Start a new conversation
Press Cmd+L (or Ctrl+L) and start fresh. Long conversations accumulate tokens.
2. Use a model with larger context
| Model | Context |
|---|---|
deepseek-v3 | 128k |
deepseek-v4-pro | 128k |
llama-3.3-70b | 128k |
llama-4-maverick | 128k |
kimi-k2.6 | 128k |
Models with 8k context (like some local Ollama models) fill up fast.
3. Reduce context
- Mention fewer files with
@ - Remove long file references from the conversation
- For autocomplete: this usually self-resolves on the next keystroke
4. For local models
If running Ollama, you can increase context with:
ollama run <model> --ctx-size 32768Or set num_ctx in your Modelfile.
Was this page helpful?
Last edited