run_1aa6fc

Scope

• Model tested: minimax/minimax-m2.5:free
• Total tests run: 7
• Categories covered: 7 — coding_ui, debugging, hallucination, reasoning, refactoring, security, speed
• Run mode: Full run (--fail-fast disabled), all tests executed
• Rate limit: 7 RPM
• Cost: Total cost \$0.00 (all completions free)

Performance

• Overall pass rate: 5/7 (71.4%)
• Total score: 5.15/7 (73.6%)
• Average latency: 34.8s
• Median latency: 11.2s
• Average TTFT (Time to First Token): 22.4s
• Average output speed: 767.0 tok/s

The high average latency is skewed by the coding_ui test (137.1s), while most other tests complete in under 12s. TTFT varies significantly across categories, from 2.7s in hallucination to 90.4s in coding_ui.

Category-Level Results

• coding_ui:

  • Passed: 0/1
  • Score: 0.15/1 (15%)
  • Latency: 137.1s (highest of all tests)
  • TTFT: 90.4s
  • Output speed: 112.5 tok/s (lowest observed)
  • Failure reason: Model generated a partial UI with excessive boilerplate but failed core functionality

• debugging, hallucination, reasoning, security, speed:

  • All passed (1/1 in each)
  • Perfect scores: 1.0 in all five
  • Fast TTFT: All under 28s, fastest at 2.7s (hallucination)
  • High output speed: Up to 1.7k tok/s (reasoning, debugging)

• refactoring:

  • Passed: 0/1
  • Score: 0/1
  • Latency: 12.4s
  • TTFT: 11.5s
  • Output speed: 542.0 tok/s
  • Failure reason: Output used list comprehension but did not match expected format or logic exactly

Notable Observations

• Extreme TTFT in coding_ui: 90.4s suggests potential inefficiency or blocking behavior when generating UI code
• Low output speed in security: Only 9.1 tok/s despite correct, detailed answer — likely due to deliberate, verbose explanation
• Zero cost across all tests: Confirmed no token-based charges for this model endpoint
• Token totals:
  • Prompt tokens: 763
  • Completion tokens: 6,976
  • Total: 7,739 tokens processed

Despite two failures, the model demonstrates strong reasoning, security awareness, and speed in most domains. The coding_ui and refactoring failures suggest weaknesses in code generation fidelity under specific patterns.