Using openrouter r1 as architect, sonnet as editor and deepseek-chat as weak-model

[GratefulDave]

Issue

I am trying to use:

• Architect: r1
• Editor: claude-3.5-sonnet
• Weak Model: deepseek-chat

When I run aider, the output is:

Found OPENROUTER_API_KEY so using openrouter/anthropic/claude-3.5-sonnet since no --model was specified.
Aider v0.73.0
Model: openrouter/anthropic/claude-3.5-sonnet with whole edit format, infinite output
Git repo: .git with 166 files
Repo-map: using 4092 tokens, auto refresh

What I expect is:

If I run aider --model r1:

aider --model r1
───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Warning: deepseek/deepseek-reasoner expects these environment variables
- DEEPSEEK_API_KEY: Not set
Warning: deepseek/deepseek-chat expects these environment variables
- DEEPSEEK_API_KEY: Not set
Warning: deepseek/deepseek-chat expects these environment variables
- DEEPSEEK_API_KEY: Not set
You can skip this check with --no-show-model-warnings

https://aider.chat/docs/llms/warnings.html
Open documentation url for more info? (Y)es/(N)o/(D)on't ask again [Yes]: N                                                                        

Aider v0.73.0
Main model: deepseek/deepseek-reasoner with diff edit format, prompt cache, infinite output
Weak model: deepseek/deepseek-chat
Git repo: .git with 166 files
Repo-map: using 4092 tokens, auto refresh

Right now I can't even get a new deepseek api token because I get a 503 every time I try to login to the api.

Here are my settings:

.aider.config.yml

api-key:
  openrouter: {api_key}
architect: true
model: openrouter/deepseek/deepseek-reasoner
editor-model: openrouter/anthropic/claude-3.5-sonnet
weak-model: openrouter/deepseek/deepseek-chat
multiline: true
watch-files: true
map-tokens: 4092
map-refresh: auto
dark-mode: true
pretty: true
cache-prompts: true

.aider.model.settings.yml

- name: openrouter/anthropic/claude-3.5-sonnet
  extra_params:
    extra_body:
      provider:
        # Only use these providers, in this order
        order: ["Anthropic"]
        # Don't fall back to other providers
        allow_fallbacks: false
        # Skip providers that may train on inputs
        data_collection: "deny"
        # Only use providers supporting all parameters
        require_parameters: true

- name: openrouter/deepseek/deepseek-r1:free
  extra_params:
    extra_body:
      provider:
        # Only use these providers, in this order
        order: ["Chute"]
        # Don't fall back to other providers
        allow_fallbacks: false
        # Skip providers that may train on inputs
        data_collection: "deny"
        # Only use providers supporting all parameters
        require_parameters: true

- name: openrouter/deepseek/deepseek-chat
  extra_params:
    extra_body:
      provider:
        # Only use these providers, in this order
        order: ["DeepSeek", "Nebia", "Fireworks"]
        # Don't fall back to other providers
        allow_fallbacks: false
        # Skip providers that may train on inputs
        data_collection:
        # Only use providers supporting all parameters
        require_parameters: true

Version and model info

Aider v0.73.0
Model: As above
Repo-Map: 4192

[V4G4X]

You have a working OpenRouter account, so why not use the Deepseek models through OpenRouter?

You have the added benefit of optionally filtering out providers that train on your input data (like DeepSeek themselves).

[akaihola]

.aider.config.yml

api-key:
openrouter: {api_key}

For me, this format works:

api-key:
- gemini={api_key}
- fireworks={api_key}

R1 is streaming and much faster overall through Fireworks AI, but it constantly hits its token limit:

Model fireworks_ai/accounts/fireworks/models/deepseek-r1 has hit a token limit!
Token counts below are approximate.

Input tokens: ~9,441 of 0 -- possibly exhausted context window!
Output tokens: ~1,908 of 0 -- possibly exceeded output limit!
Total tokens: ~11,349 of 0 -- possibly exhausted context window!

To reduce output tokens:
- Ask for smaller changes in each request.
- Break your code into smaller source files.
- Use a stronger model that can return diffs.

To reduce input tokens:
- Use /tokens to see token usage.
- Use /drop to remove unneeded files from the chat session.
- Use /clear to clear the chat history.
- Break your code into smaller source files.