Add Qwen3.6-27B-FP8 results for all 3 benchmarks

9 result files across SWE-bench Verified, SWE-bench Pro Ansible,
and RH SWE-bench with Claude Code, OpenCode, and Pi harnesses.

results/qwen3-6-27b-fp8-rh-swe-bench-claude-code.json

@@ -0,0 +1,48 @@
+{
+    "benchmark": {
+        "name": "RH SWE-Bench",
+        "repo": "rounakbende/rh-swe-bench",
+        "num_tasks": 357,
+        "url": "https://huggingface.co/datasets/rounakbende/rh-swe-bench"
+    },
+    "harness": {
+        "name": "Claude Code",
+        "skills": [],
+        "is_oss": false,
+        "url": "https://github.com/anthropics/claude-code"
+    },
+    "model": {
+        "name": "Qwen3.6-27B-FP8",
+        "repo": "RedHatAI/Qwen3.6-27B-FP8",
+        "is_oss": true,
+        "num_params": 27,
+        "precision": "fp8",
+        "url": "https://huggingface.co/RedHatAI/Qwen3.6-27B-FP8"
+    },
+    "environment": {
+        "name": "harbor",
+        "config": {
+            "name": "rh-swe-bench"
+        },
+        "url": "https://github.com/harbor-framework/harbor"
+    },
+    "metrics": {
+        "n_tasks": 357,
+        "n_errors": 9,
+        "score": 0.493,
+        "n_input_tokens": 885869433,
+        "n_cache_tokens": 0,
+        "n_output_tokens": 4763916,
+        "n_total_tokens": 890633349,
+        "agent_time_seconds": 201374,
+        "total_time_seconds": 401657,
+        "cost_usd": 73.84,
+        "mean_input_tokens_per_task": 2481426,
+        "mean_cache_tokens_per_task": 0,
+        "mean_output_tokens_per_task": 13344,
+        "mean_tokens_per_task": 2494771,
+        "mean_cost_usd_per_task": 0.21,
+        "mean_total_time_seconds_per_task": 1125,
+        "mean_agent_time_seconds_per_task": 564
+    }
+}

results/qwen3-6-27b-fp8-rh-swe-bench-opencode.json

@@ -0,0 +1,48 @@
+{
+    "benchmark": {
+        "name": "RH SWE-Bench",
+        "repo": "rounakbende/rh-swe-bench",
+        "num_tasks": 357,
+        "url": "https://huggingface.co/datasets/rounakbende/rh-swe-bench"
+    },
+    "harness": {
+        "name": "OpenCode",
+        "skills": [],
+        "is_oss": true,
+        "url": "https://github.com/opencode-ai/opencode"
+    },
+    "model": {
+        "name": "Qwen3.6-27B-FP8",
+        "repo": "RedHatAI/Qwen3.6-27B-FP8",
+        "is_oss": true,
+        "num_params": 27,
+        "precision": "fp8",
+        "url": "https://huggingface.co/RedHatAI/Qwen3.6-27B-FP8"
+    },
+    "environment": {
+        "name": "harbor",
+        "config": {
+            "name": "rh-swe-bench"
+        },
+        "url": "https://github.com/harbor-framework/harbor"
+    },
+    "metrics": {
+        "n_tasks": 357,
+        "n_errors": 10,
+        "score": 0.44,
+        "n_input_tokens": 410001939,
+        "n_cache_tokens": 0,
+        "n_output_tokens": 2851571,
+        "n_total_tokens": 412853510,
+        "agent_time_seconds": 115135,
+        "total_time_seconds": 315451,
+        "cost_usd": 42.22,
+        "mean_input_tokens_per_task": 1148464,
+        "mean_cache_tokens_per_task": 0,
+        "mean_output_tokens_per_task": 7987,
+        "mean_tokens_per_task": 1156452,
+        "mean_cost_usd_per_task": 0.12,
+        "mean_total_time_seconds_per_task": 883,
+        "mean_agent_time_seconds_per_task": 322
+    }
+}

results/qwen3-6-27b-fp8-rh-swe-bench-pi.json

@@ -0,0 +1,48 @@
+{
+    "benchmark": {
+        "name": "RH SWE-Bench",
+        "repo": "rounakbende/rh-swe-bench",
+        "num_tasks": 357,
+        "url": "https://huggingface.co/datasets/rounakbende/rh-swe-bench"
+    },
+    "harness": {
+        "name": "Pi",
+        "skills": [],
+        "is_oss": true,
+        "url": "https://github.com/plandex-ai/plandex"
+    },
+    "model": {
+        "name": "Qwen3.6-27B-FP8",
+        "repo": "RedHatAI/Qwen3.6-27B-FP8",
+        "is_oss": true,
+        "num_params": 27,
+        "precision": "fp8",
+        "url": "https://huggingface.co/RedHatAI/Qwen3.6-27B-FP8"
+    },
+    "environment": {
+        "name": "harbor",
+        "config": {
+            "name": "rh-swe-bench"
+        },
+        "url": "https://github.com/harbor-framework/harbor"
+    },
+    "metrics": {
+        "n_tasks": 357,
+        "n_errors": 1,
+        "score": 0.468,
+        "n_input_tokens": 560672462,
+        "n_cache_tokens": 0,
+        "n_output_tokens": 3618673,
+        "n_total_tokens": 564291135,
+        "agent_time_seconds": 114791,
+        "total_time_seconds": 333385,
+        "cost_usd": 42.09,
+        "mean_input_tokens_per_task": 1570511,
+        "mean_cache_tokens_per_task": 0,
+        "mean_output_tokens_per_task": 10136,
+        "mean_tokens_per_task": 1580647,
+        "mean_cost_usd_per_task": 0.12,
+        "mean_total_time_seconds_per_task": 933,
+        "mean_agent_time_seconds_per_task": 321
+    }
+}

results/qwen3-6-27b-fp8-swebench-pro-ansible-claude-code.json

@@ -0,0 +1,48 @@
+{
+    "benchmark": {
+        "name": "SWE-Bench Pro -- Ansible",
+        "repo": "scale-ai/swe-bench-pro",
+        "num_tasks": 96,
+        "url": "https://huggingface.co/datasets/scale-ai/swe-bench-pro"
+    },
+    "harness": {
+        "name": "Claude Code",
+        "skills": [],
+        "is_oss": false,
+        "url": "https://github.com/anthropics/claude-code"
+    },
+    "model": {
+        "name": "Qwen3.6-27B-FP8",
+        "repo": "RedHatAI/Qwen3.6-27B-FP8",
+        "is_oss": true,
+        "num_params": 27,
+        "precision": "fp8",
+        "url": "https://huggingface.co/RedHatAI/Qwen3.6-27B-FP8"
+    },
+    "environment": {
+        "name": "harbor",
+        "config": {
+            "name": "swe-bench-pro-ansible"
+        },
+        "url": "https://github.com/harbor-framework/harbor"
+    },
+    "metrics": {
+        "n_tasks": 96,
+        "n_errors": 8,
+        "score": 0.521,
+        "n_input_tokens": 386929017,
+        "n_cache_tokens": 0,
+        "n_output_tokens": 1847087,
+        "n_total_tokens": 388776104,
+        "agent_time_seconds": 118693,
+        "total_time_seconds": 137736,
+        "cost_usd": 43.52,
+        "mean_input_tokens_per_task": 4030510,
+        "mean_cache_tokens_per_task": 0,
+        "mean_output_tokens_per_task": 19240,
+        "mean_tokens_per_task": 4049751,
+        "mean_cost_usd_per_task": 0.45,
+        "mean_total_time_seconds_per_task": 1434,
+        "mean_agent_time_seconds_per_task": 1236
+    }
+}

results/qwen3-6-27b-fp8-swebench-pro-ansible-opencode.json

@@ -0,0 +1,48 @@
+{
+    "benchmark": {
+        "name": "SWE-Bench Pro -- Ansible",
+        "repo": "scale-ai/swe-bench-pro",
+        "num_tasks": 96,
+        "url": "https://huggingface.co/datasets/scale-ai/swe-bench-pro"
+    },
+    "harness": {
+        "name": "OpenCode",
+        "skills": [],
+        "is_oss": true,
+        "url": "https://github.com/opencode-ai/opencode"
+    },
+    "model": {
+        "name": "Qwen3.6-27B-FP8",
+        "repo": "RedHatAI/Qwen3.6-27B-FP8",
+        "is_oss": true,
+        "num_params": 27,
+        "precision": "fp8",
+        "url": "https://huggingface.co/RedHatAI/Qwen3.6-27B-FP8"
+    },
+    "environment": {
+        "name": "harbor",
+        "config": {
+            "name": "swe-bench-pro-ansible"
+        },
+        "url": "https://github.com/harbor-framework/harbor"
+    },
+    "metrics": {
+        "n_tasks": 96,
+        "n_errors": 0,
+        "score": 0.573,
+        "n_input_tokens": 344853232,
+        "n_cache_tokens": 0,
+        "n_output_tokens": 1661197,
+        "n_total_tokens": 346514429,
+        "agent_time_seconds": 118356,
+        "total_time_seconds": 196235,
+        "cost_usd": 43.4,
+        "mean_input_tokens_per_task": 3592221,
+        "mean_cache_tokens_per_task": 0,
+        "mean_output_tokens_per_task": 17304,
+        "mean_tokens_per_task": 3609525,
+        "mean_cost_usd_per_task": 0.45,
+        "mean_total_time_seconds_per_task": 2044,
+        "mean_agent_time_seconds_per_task": 1232
+    }
+}

results/qwen3-6-27b-fp8-swebench-pro-ansible-pi.json

@@ -0,0 +1,48 @@
+{
+    "benchmark": {
+        "name": "SWE-Bench Pro -- Ansible",
+        "repo": "scale-ai/swe-bench-pro",
+        "num_tasks": 96,
+        "url": "https://huggingface.co/datasets/scale-ai/swe-bench-pro"
+    },
+    "harness": {
+        "name": "Pi",
+        "skills": [],
+        "is_oss": true,
+        "url": "https://github.com/plandex-ai/plandex"
+    },
+    "model": {
+        "name": "Qwen3.6-27B-FP8",
+        "repo": "RedHatAI/Qwen3.6-27B-FP8",
+        "is_oss": true,
+        "num_params": 27,
+        "precision": "fp8",
+        "url": "https://huggingface.co/RedHatAI/Qwen3.6-27B-FP8"
+    },
+    "environment": {
+        "name": "harbor",
+        "config": {
+            "name": "swe-bench-pro-ansible"
+        },
+        "url": "https://github.com/harbor-framework/harbor"
+    },
+    "metrics": {
+        "n_tasks": 96,
+        "n_errors": 0,
+        "score": 0.49,
+        "n_input_tokens": 395189021,
+        "n_cache_tokens": 0,
+        "n_output_tokens": 2206097,
+        "n_total_tokens": 397395118,
+        "agent_time_seconds": 80305,
+        "total_time_seconds": 91995,
+        "cost_usd": 29.45,
+        "mean_input_tokens_per_task": 4116552,
+        "mean_cache_tokens_per_task": 0,
+        "mean_output_tokens_per_task": 22980,
+        "mean_tokens_per_task": 4139532,
+        "mean_cost_usd_per_task": 0.31,
+        "mean_total_time_seconds_per_task": 958,
+        "mean_agent_time_seconds_per_task": 836
+    }
+}

results/qwen3-6-27b-fp8-swebench-verified-claude-code.json

@@ -0,0 +1,48 @@
+{
+    "benchmark": {
+        "name": "SWE-Bench Verified",
+        "repo": "SWE-bench/SWE-bench_Verified",
+        "num_tasks": 500,
+        "url": "https://huggingface.co/datasets/SWE-bench/SWE-bench_Verified"
+    },
+    "harness": {
+        "name": "Claude Code",
+        "skills": [],
+        "is_oss": false,
+        "url": "https://github.com/anthropics/claude-code"
+    },
+    "model": {
+        "name": "Qwen3.6-27B-FP8",
+        "repo": "RedHatAI/Qwen3.6-27B-FP8",
+        "is_oss": true,
+        "num_params": 27,
+        "precision": "fp8",
+        "url": "https://huggingface.co/RedHatAI/Qwen3.6-27B-FP8"
+    },
+    "environment": {
+        "name": "harbor",
+        "config": {
+            "name": "swe-bench-verified"
+        },
+        "url": "https://github.com/harbor-framework/harbor"
+    },
+    "metrics": {
+        "n_tasks": 500,
+        "n_errors": 3,
+        "score": 0.694,
+        "n_input_tokens": 1005844088,
+        "n_cache_tokens": 0,
+        "n_output_tokens": 6255800,
+        "n_total_tokens": 1012099888,
+        "agent_time_seconds": 228858,
+        "total_time_seconds": 411348,
+        "cost_usd": 83.91,
+        "mean_input_tokens_per_task": 2011688,
+        "mean_cache_tokens_per_task": 0,
+        "mean_output_tokens_per_task": 12511,
+        "mean_tokens_per_task": 2024199,
+        "mean_cost_usd_per_task": 0.17,
+        "mean_total_time_seconds_per_task": 822,
+        "mean_agent_time_seconds_per_task": 457
+    }
+}

results/qwen3-6-27b-fp8-swebench-verified-opencode.json

@@ -0,0 +1,48 @@
+{
+    "benchmark": {
+        "name": "SWE-Bench Verified",
+        "repo": "SWE-bench/SWE-bench_Verified",
+        "num_tasks": 500,
+        "url": "https://huggingface.co/datasets/SWE-bench/SWE-bench_Verified"
+    },
+    "harness": {
+        "name": "OpenCode",
+        "skills": [],
+        "is_oss": true,
+        "url": "https://github.com/opencode-ai/opencode"
+    },
+    "model": {
+        "name": "Qwen3.6-27B-FP8",
+        "repo": "RedHatAI/Qwen3.6-27B-FP8",
+        "is_oss": true,
+        "num_params": 27,
+        "precision": "fp8",
+        "url": "https://huggingface.co/RedHatAI/Qwen3.6-27B-FP8"
+    },
+    "environment": {
+        "name": "harbor",
+        "config": {
+            "name": "swe-bench-verified"
+        },
+        "url": "https://github.com/harbor-framework/harbor"
+    },
+    "metrics": {
+        "n_tasks": 500,
+        "n_errors": 8,
+        "score": 0.642,
+        "n_input_tokens": 395291929,
+        "n_cache_tokens": 0,
+        "n_output_tokens": 3296668,
+        "n_total_tokens": 398588597,
+        "agent_time_seconds": 92700,
+        "total_time_seconds": 374913,
+        "cost_usd": 33.99,
+        "mean_input_tokens_per_task": 790583,
+        "mean_cache_tokens_per_task": 0,
+        "mean_output_tokens_per_task": 6593,
+        "mean_tokens_per_task": 797177,
+        "mean_cost_usd_per_task": 0.07,
+        "mean_total_time_seconds_per_task": 749,
+        "mean_agent_time_seconds_per_task": 185
+    }
+}

results/qwen3-6-27b-fp8-swebench-verified-pi.json

@@ -0,0 +1,48 @@
+{
+    "benchmark": {
+        "name": "SWE-Bench Verified",
+        "repo": "SWE-bench/SWE-bench_Verified",
+        "num_tasks": 500,
+        "url": "https://huggingface.co/datasets/SWE-bench/SWE-bench_Verified"
+    },
+    "harness": {
+        "name": "Pi",
+        "skills": [],
+        "is_oss": true,
+        "url": "https://github.com/plandex-ai/plandex"
+    },
+    "model": {
+        "name": "Qwen3.6-27B-FP8",
+        "repo": "RedHatAI/Qwen3.6-27B-FP8",
+        "is_oss": true,
+        "num_params": 27,
+        "precision": "fp8",
+        "url": "https://huggingface.co/RedHatAI/Qwen3.6-27B-FP8"
+    },
+    "environment": {
+        "name": "harbor",
+        "config": {
+            "name": "swe-bench-verified"
+        },
+        "url": "https://github.com/harbor-framework/harbor"
+    },
+    "metrics": {
+        "n_tasks": 500,
+        "n_errors": 7,
+        "score": 0.694,
+        "n_input_tokens": 544070591,
+        "n_cache_tokens": 0,
+        "n_output_tokens": 4965652,
+        "n_total_tokens": 549036243,
+        "agent_time_seconds": 130591,
+        "total_time_seconds": 408706,
+        "cost_usd": 47.88,
+        "mean_input_tokens_per_task": 1088141,
+        "mean_cache_tokens_per_task": 0,
+        "mean_output_tokens_per_task": 9931,
+        "mean_tokens_per_task": 1098072,
+        "mean_cost_usd_per_task": 0.1,
+        "mean_total_time_seconds_per_task": 817,
+        "mean_agent_time_seconds_per_task": 261
+    }
+}