Improves custom-provider model setup in DeepSeek Harness by letting each model declare its native capabilities: vision input, reasoning-effort levels, and token capacities such as context window and max output. Configuration lives in a Settings section with zero runtime changes to the harness.
Per-model capability declaration for DeepSeek Harness: reasoning-effort levels (wire spellings) + request modalities (vision) for OpenAI-compatible providers. Settings section, zero runtime harness deps, no YAML.
The official Models page edits a model's input types and token capacities, but it does so by rewriting that route's whole models array: one field touched, and every other model on the route stops following the catalog — and sparse modelOverrides becomes illegal there.
better-model-provider is a DeepSeek Harness ecosystem resource maintained by sanshanya. Improves custom-provider model setup in DeepSeek Harness by letting each model declare its native capabilities: vision input, reasoning-effort levels, and token capacities such as context window and max output. Configuration lives in a Settings section with zero runtime changes to the harness.
Source code and usage instructions are available at https://github.com/sanshanya/better-model-provider. Follow the repository README for the correct setup steps.
better-model-provider is a community open-source project released under the MIT license. Review the repository license and documentation before use.
Models & Providers
An AI SDK provider that exposes a DeepSeek Harness (dsh) runtime as a LanguageModelV3, compatible with AI SDK v6 and v7. Any application using the AI SDK can then call on dsh's full agent loop, tools, skills, MCP, and sessions as a language model.
View details → 
krislavten
Models & Providers
Harness engineering beginner tutorial, from 0 to 1
View details → 
walkinglabs
Models & Providers
An LLM-powered multi-agent framework for simulating interactive, evolving story worlds.
View details → 
ZJU-LLMs