Add swe-bench results for MiniMax-M2.7

[all-hands-bot]

Evaluation Results

Model: MiniMax-M2.7
Benchmark: swe-bench
Agent Version: v1.14.0

Results

• Accuracy: 75.6%
• Total Cost: $0.00
• Average Instance Cost: $0.00
• Total Duration: 264558s (4409.3m)
• Average Instance Runtime: 529s

---

Report Summary

• Total instances: 500
• Submitted instances: 500
• Resolved instances: 378
• Unresolved instances: 121
• Empty patch instances: 1
• Error instances: 0

Additional Metadata

• completed_instances: 499
• schema_version: 2
• unstopped_instances: 0

---

This PR was automatically created by the evaluation pipeline.

[github-actions]

📊 Progress Report

============================================================
OpenHands Index Results - Progress Report
============================================================

Target: Complete all model × benchmark pairs
  22 models × 5 benchmarks = 110 pairs
  (each pair requires all 3 metrics: score, cost_per_instance, average_runtime)

Incomplete Pairs (10):
  GPT-5.4:
    - swt-bench (all metrics)
  Qwen3.5-Flash:
    - swe-bench (all metrics)
    - swe-bench-multimodal (all metrics)
    - swt-bench (all metrics)
    - gaia (all metrics)
  Qwen3-Coder-Next:
    - swt-bench (all metrics)
  Minimax-2.7:
    - swe-bench-multimodal (all metrics)
    - swt-bench (all metrics)
    - commit0 (all metrics)
    - gaia (all metrics)

============================================================
OVERALL PROGRESS: ⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛⬜ 90.91%
  Complete: 100 / 110 pairs
============================================================

✅ Schema Validation

============================================================
Schema Validation Report
============================================================

Results directory: /home/runner/work/openhands-index-results/openhands-index-results/results
Files validated: 46
  Passed: 46
  Failed: 0

============================================================
VALIDATION PASSED
============================================================

This report measures progress towards the 3D array goal (benchmarks × models × metrics) as described in #2.

[juanmichelini]

Conversation Error Report

Archive: https://results.eval.all-hands.dev/swebench/litellm_proxy-minimax-MiniMax-M2-7/23463806447/results.tar.gz
Generated: 2026-03-24 01:31:13 UTC

Summary

Total conversations: 499
Conversations with errors: 9
Error rate: 1.8%

Error Occurrences (sorted by count)

Count % Error

---

 3    0.6%  Tool 'grep' not found. Available: ['terminal', 'file_editor', 'task_tracker', 'finish', 'think']
 1    0.2%  Failed to process archive: not a gzip file
 1    0.2%  Error executing tool 'terminal': Cannot use reset=True with is_input=True
 1    0.2%  Error validating args {"command": "view", "path": "/workspace/scikit-learn/sklearn/model_selection/_split.py", "view_range": [1167, 1225], "security_rity": "LOW", "summary": "View RepeatedKFold init"}...
 1    0.2%  MaxIterationsReached
 1    0.2%  Error validating args {"thought": "Let me understand the issue better. In 5.1.1:\n\nIn `_getconftestmodules`:\n- Old code used `directory.realpath().parts()` to iterate\n- It passed `conftestpath.real...
 1    0.2%  Error validating args {"command": "str_replace", "old_str": "# Expected: __in with [0] should be equivalent to __key = 0\nprint(\"\\n--- Verification ---\")\nif exact_items.count() != in_items.count()...

Unique error types: 7

Orchestrator / Runtime Failures (from instance logs)

Instances that required retries or failed at the harness level.
has-conv = conversation archive exists; no-conv = completely failed.

Instance Retries Status Error

---

sphinx-doc__sphinx-9320 3 has-conv runtime_failure_count=3
django__django-11885 1 has-conv runtime_failure_count=1
matplotlib__matplotlib-26208 1 has-conv runtime_failure_count=1
sphinx-doc__sphinx-8056 1 has-conv runtime_failure_count=1

Instances with runtime failures: 4

• with conversation archive: 4
• without conversation archive: 0