Skip to content

Commit 6e7d089

Browse files
author
zhenjun.chen
committed
docs: add OrcaRouter documentation
Documentation for the OrcaRouter provider with usage examples covering: - Quickstart YAML and JSON config - Pinning specific upstream models - Reasoning controls (OpenAI reasoning_effort, Anthropic thinking block, Gemini caveat) - extra_body fallback chain - Agent / tool-calling caveat for orcarouter/auto - Anthropic prompt caching Also adds the OrcaRouter row to the providers overview Hosted Services table and registers the page in the docs sidebar.
1 parent 1ab4ec6 commit 6e7d089

3 files changed

Lines changed: 197 additions & 0 deletions

File tree

‎docs/customize/model-providers/more/orcarouter.mdx‎

Lines changed: 195 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,195 @@
1+
---
2+
title: "How to Configure OrcaRouter with Continue"
3+
sidebarTitle: "OrcaRouter"
4+
---
5+
6+
<Info>
7+
[OrcaRouter](https://www.orcarouter.ai) is an OpenAI-compatible API gateway that aggregates ~120 chat models from OpenAI, Anthropic, Google, DeepSeek, xAI, Qwen, Kimi, MiniMax, Z-AI, and other vendors behind a single `sk-orca-` key.
8+
</Info>
9+
10+
<Tip>
11+
Sign up at [orcarouter.ai](https://www.orcarouter.ai) and obtain an API key from your [console](https://www.orcarouter.ai/console).
12+
</Tip>
13+
14+
## Quickstart
15+
16+
<Tabs>
17+
<Tab title="YAML">
18+
```yaml title="config.yaml"
19+
name: My Config
20+
version: 0.0.1
21+
schema: v1
22+
23+
models:
24+
- name: OrcaRouter Auto
25+
provider: orcarouter
26+
model: orcarouter/auto
27+
apiBase: https://api.orcarouter.ai/v1/
28+
apiKey: sk-orca-xxxxxxxxxxxxxxxxxx
29+
```
30+
</Tab>
31+
<Tab title="JSON (Deprecated)">
32+
```json title="config.json"
33+
{
34+
"models": [
35+
{
36+
"title": "OrcaRouter Auto",
37+
"provider": "orcarouter",
38+
"model": "orcarouter/auto",
39+
"apiBase": "https://api.orcarouter.ai/v1/",
40+
"apiKey": "sk-orca-xxxxxxxxxxxxxxxxxx"
41+
}
42+
]
43+
}
44+
```
45+
</Tab>
46+
</Tabs>
47+
48+
`orcarouter/auto` is a virtual model that adaptively routes each request to a candidate upstream based on a configurable strategy (cheapest / balanced / quality / contextual bandit / difficulty-gated). Routing pools and reward weights are tunable from the [console](https://www.orcarouter.ai/console/routing) without changing client code.
49+
50+
## Pinning a specific model
51+
52+
You can pin any model from the [OrcaRouter catalog](https://www.orcarouter.ai/models) by passing its full ID:
53+
54+
<Tabs>
55+
<Tab title="YAML">
56+
```yaml title="config.yaml"
57+
models:
58+
- name: Claude Opus 4.7
59+
provider: orcarouter
60+
model: anthropic/claude-opus-4.7
61+
apiBase: https://api.orcarouter.ai/v1/
62+
apiKey: ${{ secrets.ORCAROUTER_API_KEY }}
63+
64+
- name: GPT-5.5
65+
provider: orcarouter
66+
model: openai/gpt-5.5
67+
apiBase: https://api.orcarouter.ai/v1/
68+
apiKey: ${{ secrets.ORCAROUTER_API_KEY }}
69+
70+
- name: DeepSeek V4 Pro
71+
provider: orcarouter
72+
model: deepseek/deepseek-v4-pro
73+
apiBase: https://api.orcarouter.ai/v1/
74+
apiKey: ${{ secrets.ORCAROUTER_API_KEY }}
75+
```
76+
</Tab>
77+
</Tabs>
78+
79+
## Reasoning controls
80+
81+
OrcaRouter passes vendor-native reasoning controls through to the upstream:
82+
83+
### OpenAI / Grok / Gemini / Qwen reasoning families
84+
85+
Use the flat `reasoning_effort` field via `requestOptions.extraBodyProperties`:
86+
87+
```yaml title="config.yaml"
88+
models:
89+
- name: GPT-5.5 (High Reasoning)
90+
provider: orcarouter
91+
model: openai/gpt-5.5
92+
apiBase: https://api.orcarouter.ai/v1/
93+
apiKey: ${{ secrets.ORCAROUTER_API_KEY }}
94+
requestOptions:
95+
extraBodyProperties:
96+
reasoning_effort: high # "minimal" | "low" | "medium" | "high"
97+
```
98+
99+
### Gemini reasoning caveat
100+
101+
Some Gemini models (including `gemini-3-flash-preview`) are reasoning models that spend most of their `completion_tokens` budget on internal reasoning before producing the final reply. The streaming response stays silent for several seconds during that period, which can look like the chat is blank or stuck.
102+
103+
For fast responses, pass `reasoning_effort: "minimal"`:
104+
105+
```yaml title="config.yaml"
106+
models:
107+
- name: Gemini 3 Flash (Fast)
108+
provider: orcarouter
109+
model: google/gemini-3-flash-preview
110+
apiBase: https://api.orcarouter.ai/v1/
111+
apiKey: ${{ secrets.ORCAROUTER_API_KEY }}
112+
requestOptions:
113+
extraBodyProperties:
114+
reasoning_effort: minimal
115+
```
116+
117+
### Anthropic Claude (thinking block)
118+
119+
Anthropic reasoning models use a native `thinking` block:
120+
121+
```yaml title="config.yaml"
122+
models:
123+
- name: Claude Opus 4.7 (Thinking)
124+
provider: orcarouter
125+
model: anthropic/claude-opus-4.7
126+
apiBase: https://api.orcarouter.ai/v1/
127+
apiKey: ${{ secrets.ORCAROUTER_API_KEY }}
128+
requestOptions:
129+
extraBodyProperties:
130+
thinking:
131+
type: enabled
132+
budget_tokens: 2000 # >= 1024, must be < max_tokens
133+
```
134+
135+
<Note>
136+
Anthropic reasoning models (e.g. `claude-opus-4.7`) reject `temperature` and `top_k`. Omit them from `completionOptions` when targeting these models.
137+
</Note>
138+
139+
## Fallback chain
140+
141+
When the primary upstream fails, OrcaRouter can fall back to a configured list using the `extra_body` field:
142+
143+
```yaml title="config.yaml"
144+
models:
145+
- name: OrcaRouter Auto with Fallback
146+
provider: orcarouter
147+
model: orcarouter/auto
148+
apiBase: https://api.orcarouter.ai/v1/
149+
apiKey: ${{ secrets.ORCAROUTER_API_KEY }}
150+
requestOptions:
151+
extraBodyProperties:
152+
extra_body:
153+
models: [deepseek/deepseek-v4-pro]
154+
route: fallback
155+
```
156+
157+
## Agent / tool-calling caveat
158+
159+
The default `orcarouter/auto` pool may include models that do not support function calling. If you use Continue's Agent mode (which sends a `tools` field), pin a specific tool-capable model like `anthropic/claude-opus-4.7` or `openai/gpt-5.5`, or adjust the AUTO routing pool from the [console](https://www.orcarouter.ai/console/routing) to only include tool-capable upstreams.
160+
161+
```yaml title="config.yaml"
162+
models:
163+
- name: Claude Opus 4.7 (for Agent)
164+
provider: orcarouter
165+
model: anthropic/claude-opus-4.7
166+
apiBase: https://api.orcarouter.ai/v1/
167+
apiKey: ${{ secrets.ORCAROUTER_API_KEY }}
168+
capabilities:
169+
- tool_use
170+
roles:
171+
- chat
172+
- edit
173+
```
174+
175+
## Prompt caching
176+
177+
For Anthropic Claude models, Continue automatically injects `cache_control: { type: "ephemeral" }` on the system message and the last two user turns when `cacheBehavior` or `promptCaching` is enabled:
178+
179+
```yaml title="config.yaml"
180+
models:
181+
- name: Claude Opus 4.7 (Cached)
182+
provider: orcarouter
183+
model: anthropic/claude-opus-4.7
184+
apiBase: https://api.orcarouter.ai/v1/
185+
apiKey: ${{ secrets.ORCAROUTER_API_KEY }}
186+
cacheBehavior:
187+
cacheSystemMessage: true
188+
cacheConversation: true
189+
```
190+
191+
## See also
192+
193+
- [Full model catalog](https://www.orcarouter.ai/models)
194+
- [Routing configuration console](https://www.orcarouter.ai/console/routing)
195+
- [API documentation](https://docs.orcarouter.ai)

‎docs/customize/model-providers/overview.mdx‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -34,6 +34,7 @@ Beyond the top-level providers, Continue supports many other options:
3434
| [Together AI](/customize/model-providers/more/together) | Platform for running a variety of open models |
3535
| [DeepInfra](/customize/model-providers/more/deepinfra) | Hosting for various open source models |
3636
| [OpenRouter](/customize/model-providers/top-level/openrouter) | Gateway to multiple model providers |
37+
| [OrcaRouter](/customize/model-providers/more/orcarouter) | OpenAI-compatible gateway aggregating ~120 chat models with adaptive routing |
3738
| [ClawRouter](/customize/model-providers/more/clawrouter) | Open-source LLM router with automatic cost-optimized model selection |
3839
| [Tetrate Agent Router Service](/customize/model-providers/top-level/tetrate_agent_router_service) | Gateway with intelligent routing across multiple model providers |
3940
| [Cohere](/customize/model-providers/more/cohere) | Models specialized for semantic search and text generation |

‎docs/docs.json‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -180,6 +180,7 @@
180180
"customize/model-providers/more/moonshot",
181181
"customize/model-providers/more/nous",
182182
"customize/model-providers/more/nvidia",
183+
"customize/model-providers/more/orcarouter",
183184
"customize/model-providers/more/tensorix",
184185
"customize/model-providers/more/together",
185186
"customize/model-providers/more/xAI",

0 commit comments

Comments
 (0)