polymech/zeroclaw
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
fa91b6a170
zeroclaw/scripts/README_QWEN.md
ZeroClaw Bot fa91b6a170 feat(providers): implement Qwen OAuth quota tracking 
Add static quota display for Qwen OAuth provider (portal.qwen.ai).
Qwen OAuth API does not return rate-limit headers, so this provides
a static quota indicator based on known OAuth free-tier limits.

Changes:
- Add QwenQuotaExtractor in quota_adapter.rs
  - Parses rate-limit errors for retry backoff
  - Registered for all Qwen aliases (qwen, qwen-code, dashscope, etc.)
- Add Qwen OAuth detection in quota_cli.rs
  - Auto-detects ~/.qwen/oauth_creds.json
  - Displays static quota: ?/1000 (unknown remaining, 1000/day total)
- Improve quota display formatting
  - Shows "?/total" when only total limit is known
- Add comprehensive test report and testing scripts
  - Full integration test report: docs/qwen-provider-test-report.md
  - Model availability, context window, and latency tests
  - Reusable test scripts in scripts/ directory

Test results:
- Available model: qwen3-coder-plus (verified)
- Context window: ~32K tokens
- Average latency: ~2.8s
- All 15 quota tests passing

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-02-24 10:16:37 +07:00

1.2 KiB
Raw Blame History

Qwen Provider Testing Scripts

These scripts were used for initial validation and testing of the Qwen OAuth provider integration.

Scripts

qwen_model_probe.sh

Tests availability of different Qwen models through OAuth API.

Usage:

./scripts/qwen_model_probe.sh

Output: CSV with model availability results (saved to docs/qwen_model_test_results.csv)

qwen_context_test.sh

Tests context window limits by sending progressively larger prompts.

Usage:

./scripts/qwen_context_test.sh

Tests: 1K, 2K, 4K, 8K, 16K, 32K, 65K, 131K token contexts

qwen_latency_benchmark.sh

Measures response latency for Qwen provider.

Usage:

./scripts/qwen_latency_benchmark.sh [iterations]

Default: 3 iterations

Output: Individual and average response times

Test Results

See full test report: docs/qwen-provider-test-report.md

Summary:

  • ✅ Available model: qwen3-coder-plus
  • ✅ Context window: ~32K tokens
  • ✅ Average latency: ~2.8s
  • ✅ Quota tracking: Static display (?/1000)

Date: 2026-02-24 Status: Testing complete, scripts preserved for future validation