v0.2.8

See README for usage instructions.

v0.2.7

See README for usage instructions.

v0.2.6

See README for usage instructions.

v0.2.5

See README for usage instructions.

v0.2.4

See README for usage instructions.

v0.2.3

See README for usage instructions.

v0.2.1

What's New in v0.2.1

Pre-populated Default Config

New users now see 7 providers, 6 routes, and 4 tags out of the box — just replace your-key-here with real API keys to get started.

Providers: DeepSeek (OpenAI + Anthropic), Zhipu GLM, Baidu ERNIE, Kimi, Qwen/DashScope, MiniMax

Routes:

Model	Provider	Tag
deepseek-v4-pro	DeepSeek	sonnet, auto
deepseek-v4-flash	DeepSeek	haiku
glm-5.1	Zhipu GLM	opus
qwen3.7-max	Qwen/DashScope	sonnet
K2.6	Kimi	sonnet
MiniMax-M3	MiniMax	haiku

Import/Export Config

• Export: Download current config as model-router-config.json (management_key replaced with placeholder for security)
• Import: Upload a JSON config file to restore or migrate settings

Codex Tier Routing

Codex model names now map to opus/sonnet/haiku tags, sharing the same routing rules as Claude Code:

• gpt-5.5 → opus, gpt-5.4 → sonnet, gpt-5.4-mini → haiku, gpt-5.2 → haiku

Other Improvements

• Add openai_responses format option in route editor UI
• Fix frontend API auth for takeover operations
• Auto-refresh stale takeover configs on startup when port changes

Downloads

File	Platform
Model Router_0.2.1_aarch64.dmg	macOS (Apple Silicon)
Windows/Linux builds	Added by CI automatically