Add config.yaml custom model routing with quota-aware provider selection

[Copilot]

Adds LiteLLM-style config.yaml support for defining named model routes that fan out to real providers with optional quota/error-count conditions, quota caching, and 429-triggered cache invalidation.

Core module: g4f/providers/config_provider.py

• QuotaCache – TTL-based cache (5 min) for get_quota() results per provider; immediately invalidated on 429 responses
• ErrorCounter – Rolling 1-hour window error counter per provider
• evaluate_condition() – Safe recursive-descent parser supporting provider-specific quota fields via dot-notation (e.g. quota.balance > 0, quota.credits.remaining > 0 or error_count < 3); no eval
• RouterConfig – Parses and holds routes from config.yaml
• ConfigModelProvider – AsyncGeneratorProvider that iterates configured providers, evaluates conditions, records errors, and invalidates quota cache on 429

Integration

• g4f/cookies.py – read_cookie_files() now loads config.yaml from the cookies directory alongside .har/.json files
• g4f/client/service.py – get_model_and_provider() checks RouterConfig before standard model resolution, making custom names transparent to callers

Condition variables

Each provider returns its own quota format from get_quota(). Conditions support:

Variable	Description
quota	Full quota dict from get_quota() — access any field with dot-notation
quota.balance	PollinationsAI: {"balance": float}
quota.credits.remaining	Yupp: {"credits": {"remaining": int, "total": int}}
balance	Shorthand alias for quota.balance (PollinationsAI backward compat)
error_count	Recent errors for this provider in the last hour — works universally

Missing quota keys resolve to 0.0.

Config format

Place config.yaml in your cookies dir (~/.config/g4f/cookies/ or ./har_and_cookies/):

models:
  - name: "my-gpt4"
    providers:
      - provider: "OpenaiAccount"
        model: "gpt-4o"
        condition: "balance > 0 or error_count < 3"   # PollinationsAI-style
      - provider: "PollinationsAI"   # unconditional fallback
        model: "openai-large"

  - name: "yupp-chat"
    providers:
      - provider: "Yupp"
        model: "gpt-4o"
        condition: "quota.credits.remaining > 0 or error_count < 3"  # Yupp-specific
      - provider: "PollinationsAI"
        model: "openai-large"

  - name: "llama-fast"
    providers:
      - provider: "Groq"
        model: "llama-3.3-70b"
        condition: "error_count < 3"   # universal
      - provider: "DeepInfra"
        model: "meta-llama/Llama-3.3-70B-Instruct"

Requires pyyaml (already in requirements.txt); gracefully skipped with a log warning if absent.

See docs/config-yaml-routing.md for full condition syntax reference and Python API.

Original prompt

---

This section details on the original issue you should resolve

<issue_title>add config.yaml for custom model routing</issue_title>
<issue_description>Add config.yaml support from cookies/config dir like in litelm. Support if condiotion with (if get_qutoa.balance >0 or error_count < 3. Cache qutoa result, refresh on 429. Add all wanted/needed functions and documentaion, tests</issue_description>

Comments on the Issue (you are @copilot in this section)

• Fixes add config.yaml for custom model routing #3393

---

💬 Send tasks to Copilot coding agent from Slack and Teams to turn conversations into code. Copilot posts an update in your thread when it's finished.

[hlohaus]

@copilot quota.balance is only for pollinations. All providers have his own qouta stats format

[Copilot]

@copilot quota.balance is only for pollinations. All providers have his own qouta stats format

Fixed in 8fd8cd5. evaluate_condition now accepts the full quota dict from get_quota() and exposes it as quota.* dot-notation in conditions, since each provider returns its own schema:

• PollinationsAI → {"balance": float} → quota.balance > 0 (or the balance shorthand)
• Yupp → {"credits": {"remaining": int, "total": int}} → quota.credits.remaining > 0
• PuterJS, GeminiCLI, GithubCopilot → their own formats, accessed via quota.<field>
• Missing keys resolve to 0.0 instead of raising

balance is kept as a backward-compat alias for quota.balance (PollinationsAI), and error_count < N continues to work universally for any provider.

Warning

Firewall rules blocked me from connecting to one or more addresses (expand for details)

I tried to connect to the following addresses, but was blocked by firewall rules:

• api.deepinfra.com
  • Triggering command: /usr/bin/python python -m etc.unittest (dns block)
• api.puter.com
  • Triggering command: /usr/bin/python python -m etc.unittest (dns block)
• chat.z.ai
  • Triggering command: /usr/bin/python python -m etc.unittest (dns block)
• chat3.eqing.tech
  • Triggering command: /usr/bin/python python -m etc.unittest (dns block)
• g4f.space
  • Triggering command: /usr/bin/python python -m etc.unittest (dns block)
• https://api.github.com/repos/xtekky/gpt4free/releases/latest
  • Triggering command: /usr/bin/python python -m etc.unittest (http block)
• huggingface.co
  • Triggering command: /usr/bin/python python -m etc.unittest (dns block)
• image.pollinations.ai
  • Triggering command: /usr/bin/python python -m etc.unittest (dns block)

If you need me to access, download, or install something from one of these locations, you can either:

• Configure Actions setup steps to set up my environment, which run before the firewall is enabled
• Add the appropriate URLs or hosts to the custom allowlist in this repository's Copilot coding agent settings (admins only)