Answer: Install llama.cpp from llama.app, start the server in router mode with `llama serve` (optionally passing `-hf` to download and run a specific model), then in Zed select a llama.cpp model from the model dropdown. Zed automatically discovers served models, and you can set `auto_discover` to `false` to list models yourself via `available_models`. Status: Overdue (check was due 2026-11-20T13:39:06Z and has not succeeded) Entry: tool/zed/models/configure-llamacpp-local-model Revision: 1 Digest: sha256:1a0432bda6a6508f (changes whenever the content changes) Timeline: Created: 2026-11-13T04:00:41Z Last changed: 2026-11-13T04:00:41Z Last verified: 2026-11-19T13:39:06Z Next check: 2026-11-20T13:39:06Z Check interval: Daily Expected change: No fixed schedule Data: Field | Value | Source | As of server command | llama serve | [1] | 2026-11-19 specific model command | llama serve -hf unsloth/gemma-4-26B-A4B-it-GGUF:BF16 | [1] | 2026-11-19 auto_discover setting | false | [1] | 2026-11-19 api_url | http://localhost:8080 | [1] | 2026-11-19 max_tokens | 32768 | [1] | 2026-11-19 context_window | 8192 | [1] | 2026-11-19 API key env var | LLAMACPP_API_KEY | [1] | 2026-11-19 Sources: [1] Zed Industries. "zed.dev/docs/ai/use-a-local-model.md". https://zed.dev/docs/ai/use-a-local-model.md Role: primary. Accessed: 2026-11-19T13:39:06Z Quote: "llama serve" (Supports: server command) Quote: "llama serve -hf unsloth/gemma-4-26B-A4B-it-GGUF:BF16" (Supports: specific model command) Quote: "set `auto_discover` to `false`" (Supports: auto_discover setting) Quote: ""api_url": "http://localhost:8080"" (Supports: api_url) Quote: ""max_tokens": 32768" (Supports: max_tokens) Quote: ""context_window": 8192" (Supports: context_window) Quote: "set `LLAMACPP_API_KEY`" (Supports: API key env var) Snapshot SHA-256: f6045bc52ae2d417e103465270e31b0df0bfc31d0f14e5274be0f7987abd3ed3 (captured 2026-11-13T04:00:41Z) Verification: Method: Automated extraction from [1] with deepseek-flash (JSON output, thinking disabled). Every value was checked by code to appear verbatim in a quoted passage of [1]. On this check the text of [1] was unchanged since 2026-11-13T04:00:41Z, so the values verified then were kept. Cross-check: not-available at 2026-11-19T13:39:06Z. No independent second source is configured for this entry Last successful check: 2026-11-19T13:39:06Z Last attempt: 2026-11-19T13:39:06Z Notes and limitations: Extracted automatically from a single official source (section "llama.cpp {#llama-cpp}"). Human-readable page: https://zed.dev/docs/ai/use-a-local-model.html Related: https://latestn.com/tool/zed/models/configure-llm-providers How do I configure LLM providers for Zed AI features? https://latestn.com/tool/zed/models/configure-model-access How do I configure which language model provider Zed's AI features use? https://latestn.com/tool/zed/models/configure-ollama-local-model How do I configure Ollama as a local model provider in Zed? https://latestn.com/tool/zed/models/change-agent-model How do I change the AI model used in Zed's Agent Panel? https://latestn.com/tool/zed/models/choose-agent-models How do I choose which models the Zed Agent uses? https://latestn.com/tool/zed/models/choose-model-access-path How do I choose which model access path to use for Zed's AI features? https://latestn.com/tool/ All entries in AI tools Change log: Time | Event | Change | Source 2026-11-13T04:00:41Z | added | Entry created | — Changes for this entry as a feed: https://latestn.com/tool/zed/models/configure-llamacpp-local-model/latest Challenge this entry: Open challenges: None How: If a value here is wrong or out of date, send what is wrong and a link to evidence. Challenges are queued and checked against the evidence; they do not change the entry by themselves. If the entry changes as a result, the change log says so. Form: https://latestn.com/dispute/tool/zed/models/configure-llamacpp-local-model POST: https://latestn.com/dispute/tool/zed/models/configure-llamacpp-local-model with claim, evidence_url, and optionally field and contact (form-encoded or JSON) GET: https://latestn.com/dispute/tool/zed/models/configure-llamacpp-local-model?claim={what is wrong}&evidence_url={link} How to cite: LatestN. (2026). How do I configure a llama.cpp server to use local models with Zed AI features? Revision 1, last verified 2026-11-19T13:39:06Z. https://latestn.com/tool/zed/models/configure-llamacpp-local-model (accessed 2026-11-21T10:01:19Z). @misc{LatestN_tool_zed_models_configure_llamacpp_local_model_r1, title = {How do I configure a llama.cpp server to use local models with Zed AI features?}, author = {{LatestN}}, year = {2026}, howpublished = {\url{https://latestn.com/tool/zed/models/configure-llamacpp-local-model}}, note = {Revision 1; digest sha256:1a0432bda6a6508f; last verified 2026-11-19T13:39:06Z; accessed 2026-11-21T10:01:19Z} } Reuse: License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/) Attribution: Copying and adapting are allowed, including commercially, if you credit LatestN and link to the entry you used. Each entry ends with a ready-made citation. URL: https://latestn.com/tool/zed/models/configure-llamacpp-local-model