# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("210Codelabs/AdaptIQ-7B", device_map="auto")AdaptIQ-7B
AdaptIQ is an open-weight, self-hosted routed-expert AI system from 210 Code Labs.
This repository publishes the AdaptIQ model card, manifest, and deployment metadata for
AdaptIQ-7B.
Architecture
AdaptIQ uses expert composition instead of unsafe direct weight blending across incompatible architectures:
| Capability | Expert |
|---|---|
| Text / reasoning | mistralai/Mistral-7B-Instruct-v0.3 |
| Code / debugging | ibm-granite/granite-8b-code-instruct |
| Image generation | stabilityai/sd-turbo |
Composition method: routed_experts. alpha=0.5 is retained only for compatible linear-merge
experiments and is not used by the routed-expert release path.
Capabilities
- General text chat
- Code generation and debugging
- Agent-oriented reasoning workflows
- Image generation through the routed image expert
- OpenAI-compatible hosted API on Modal
- MCP-ready tool schema scaffold
Hosted API
Modal endpoint:
https://codelabs--adaptiq-api-api.modal.run/v1
Model name:
AdaptIQ-7B
Example:
curl https://codelabs--adaptiq-api-api.modal.run/v1/models \
-H "Authorization: Bearer $ADAPTIQ_API_KEY"
Intended use
AdaptIQ is intended for software development, automation, research, education, and creative self-hosted generation workflows.
Limitations
This release is a routed expert system metadata release. It does not package all upstream expert weights into a single checkpoint. Runtime deployments download/load the configured open-weight experts according to the AdaptIQ manifest.
License
The AdaptIQ release metadata is Apache-2.0. Respect the licenses and model cards of all upstream experts before redistribution or commercial deployment.
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="210Codelabs/AdaptIQ-7B")