DaoXEgptme runs an agent loop in your terminal — it writes code, uses the shell, browses the web. Custom endpoints come through named [[providers]] entries in ~/.config/gptme/config.toml, or the local/ prefix with OPENAI_BASE_URL; either shape fits DaoXE (cheapest APIs guide).
Why route gptme through DaoXE#
- A real tool loop. gptme executes shell commands and applies edits in your workspace — DaoXE models that truly do function calling drive it end to end.
- Keys stay out of the config file.
api_key_envpoints at an environment variable name, so the secret itself never lands in config.toml. - Per-provider defaults. A
default_modelper provider means-m daoxealone is a valid selection once the block exists.
Set it up in gptme#
- Declare the provider. In
~/.config/gptme/config.tomladd[[providers]]withname = "daoxe",base_url = "https://daoxe.com/v1"(the docs' custom-provider examples all keep the/v1suffix), andapi_key_env = "DAOXE_API_KEY". - Set the key.
export DAOXE_API_KEY="sk-you…-key"— resolution order is inlineapi_key, thenapi_key_env, then${PROVIDER_NAME}_API_KEY(hereDAOXE_API_KEYwould be found either way). - Pick the model.
gptme 'hello' -m daoxe/YOUR_EXACT_MODEL_ID— the model isprovider_name/model, with the ID copied exactly fromGET /v1/models. - Test the tool loop. Run a real command, e.g.
gptme 'list the files in the current directory' -m daoxe/YOUR_EXACT_MODEL_ID— if gptme executes a shell command and shows real filenames, tool calling works end to end.
# ~/.config/gptme/config.toml — register DaoXE as a named provider
[[providers]]
name = "daoxe"
base_url = "https://daoxe.com/v1"
api_key_env = "DAOXE_API_KEY"
default_model = "YOUR_EXACT_MODEL_ID"
# then: gptme 'hello' -m daoxe/YOUR_EXACT_MODEL_IDCaveats worth knowing
Streaming is unconditional: the docs state the transport sends stream=True regardless of metadata — an endpoint that rejects streaming will fail here; DaoXE's OpenAI-compatible path accepts streamed chat completions. Model IDs are provider-prefixed: the docs call the bare-name form a pitfall — daoxe/YOUR_EXACT_MODEL_ID, never the bare ID; a name gptme doesn't recognize warns "unknown model" but still works. The local/ prefix is env-based: OPENAI_BASE_URL + OPENAI_API_KEY also work, but those variables are shared with gptme's built-in openai provider — a named block is the cleaner isolation. Token counting may phone home: gptme fetches the OpenAI cl100k_base tokenizer and can time out offline; PyPI releases lack the character-estimate fallback, so pre-cache with tiktoken if you work airgapped. More clients: client setup notes.
Verify you actually get the model#
Prove the endpoint works before blaming the client — if this fails, no setting will fix it:
export DAOXE_API_KEY="your_api_key"
# List the exact model IDs your account can call
curl --fail-with-body --show-error --silent \
https://daoxe.com/v1/models \
-H "Authorization: Bearer ${DAOXE_API_KEY}"Prove the endpoint outside the agent, then diff a hard prompt against the official API at temperature 0:
curl --fail-with-body --show-error --silent \
https://daoxe.com/v1/chat/completions \
-H "Authorization: Bearer ${DAOXE_API_KEY}" \
-H "Content-Type: application/json" \
-d '{
"model": "YOUR_EXACT_MODEL_ID",
"max_tokens": 64,
"messages": [{"role": "user", "content": "Say hello in one sentence."}]
}'Verify us — don't trust us
Point the open benchmark at DaoXE and at the official API and compare at temperature 0. Then learn to detect model swapping so a cheaper endpoint can't quietly swap you to a smaller model.
Frequently asked questions#
Can gptme use DaoXE?
Yes — a [[providers]] block with base_url = "https://daoxe.com/v1" and api_key_env = "DAOXE_API_KEY", then -m daoxe/YOUR_EXACT_MODEL_ID.
Do I need OPENAI_BASE_URL?
No for the named block — that's the local/ prefix path (OPENAI_BASE_URL + OPENAI_API_KEY). The [[providers]] entry keeps base URL and key scoped to the daoxe provider.
Why does tool calling loop or fail?
gptme's docs attribute tool-loop failures to models not following its tool protocol — pick a model that really does function calling, or try --tool-format xml. Run a command-listing test to confirm the loop executes.
What does it cost?
Per-model, account-scoped — live pricing; top-up is a flat 1 RMB = $1 of credit on every payment method.
Try DaoXE — and benchmark it yourself
One key for GPT, Claude, Gemini, DeepSeek and more. Point the open benchmark at us and compare — don't take our word for it.