Skip to content

⚡ January 27, 2026

Generated: 2026-01-27 16:11 UTC
Total Duration: 1h 28m 43s
Iterations: 1
Judge (classifier) model: gpt-4.1

Fast Benchmark

Markers: regression or benchmark
Schedule: Weekly (Sunday 2 AM UTC)
Purpose: Quick regression tests to catch breaking changes

HolmesGPT is continuously evaluated against real-world Kubernetes and cloud troubleshooting scenarios.

If you find scenarios that HolmesGPT does not perform well on, please consider adding them as evals to the benchmark.

Model Accuracy Comparison

Model Pass Fail Skip/Error Total Success Rate
deepseek/deepseek-chat 10 4 0 14 🟡 71% (10/14)
deepseek/deepseek-reasoner 10 4 0 14 🟡 71% (10/14)
eu.anthropic.claude-haiku-4-5-20251001-v1:0 11 3 0 14 🟡 79% (11/14)
eu.anthropic.claude-opus-4-5-20251101-v1:0 11 3 0 14 🟡 79% (11/14)
eu.anthropic.claude-sonnet-4-5-20250929-v1:0 10 4 0 14 🟡 71% (10/14)
gemini/gemini-3-flash-preview 10 4 0 14 🟡 71% (10/14)
gemini/gemini-3-pro-preview 7 7 0 14 🟡 50% (7/14)
gpt-5.2 9 5 0 14 🟡 64% (9/14)

Model Cost Comparison

Model Tests Avg Cost Min Cost Max Cost Total Cost
deepseek/deepseek-chat 14 $0.01 $0.00 $0.04 $0.21
deepseek/deepseek-reasoner 14 $0.02 $0.00 $0.04 $0.25
eu.anthropic.claude-haiku-4-5-20251001-v1:0 14 $0.06 $0.02 $0.11 $0.78
eu.anthropic.claude-opus-4-5-20251101-v1:0 14 $0.26 $0.10 $0.48 $3.65
eu.anthropic.claude-sonnet-4-5-20250929-v1:0 14 $0.19 $0.07 $0.29 $2.73
gemini/gemini-3-flash-preview 12 $0.05 $0.02 $0.12 $0.63
gemini/gemini-3-pro-preview 11 $0.19 $0.03 $0.31 $2.04
gpt-5.2 14 $0.11 $0.02 $0.24 $1.51

Model Latency Comparison

Model Avg (s) Min (s) Max (s) P50 (s) P95 (s)
deepseek/deepseek-chat 122.9 19.4 210.1 129.5 210.1
deepseek/deepseek-reasoner 235.2 7.8 567.0 253.2 567.0
eu.anthropic.claude-haiku-4-5-20251001-v1:0 26.7 4.9 52.2 28.0 52.2
eu.anthropic.claude-opus-4-5-20251101-v1:0 48.5 6.0 194.7 39.0 194.7
eu.anthropic.claude-sonnet-4-5-20250929-v1:0 34.5 5.0 57.7 36.0 57.7
gemini/gemini-3-flash-preview 99.3 19.6 452.8 68.4 452.8
gemini/gemini-3-pro-preview 702.3 10.4 1523.9 883.4 1523.9
gpt-5.2 39.3 12.0 73.8 39.4 73.8

Performance by Tag

Success rate by test category and model:

Tag deepseek/deepseek-chat deepseek/deepseek-reasoner eu.anthropic.claude-haiku-4-5-20251001-v1:0 eu.anthropic.claude-opus-4-5-20251101-v1:0 eu.anthropic.claude-sonnet-4-5-20250929-v1:0 gemini/gemini-3-flash-preview gemini/gemini-3-pro-preview gpt-5.2 Warnings
benchmark 🟡 80% (⅘) 🟡 60% (⅗) 🟡 80% (⅘) 🟡 60% (⅗) 🟡 40% (⅖) 🟡 80% (⅘) 🟡 80% (⅘) 🟡 60% (⅗)
context_window 🟢 100% (2/2) 🟢 100% (2/2) 🟢 100% (2/2) 🟢 100% (2/2) 🔴 0% (0/2) 🟢 100% (2/2) 🟡 50% (½) 🟢 100% (2/2)
counting 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🔴 0% (0/1) 🟢 100% (1/1)
datetime 🟡 67% (⅔) 🟢 100% (3/3) 🟢 100% (3/3) 🟢 100% (3/3) 🟡 33% (⅓) 🟡 67% (⅔) 🟡 33% (⅓) 🟡 67% (⅔)
easy 🟡 57% (4/7) 🟡 71% (5/7) 🟡 71% (5/7) 🟡 86% (6/7) 🟡 86% (6/7) 🟡 57% (4/7) 🟡 29% (2/7) 🟡 57% (4/7)
grafana-dashboard 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1)
hard 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1)
kubernetes 🟡 86% (6/7) 🟡 71% (5/7) 🟡 71% (5/7) 🟡 86% (6/7) 🟡 86% (6/7) 🟡 86% (6/7) 🟡 43% (3/7) 🟡 86% (6/7)
logs 🟡 50% (2/4) 🟡 50% (2/4) 🟡 50% (2/4) 🟡 50% (2/4) 🔴 0% (0/4) 🟡 75% (¾) 🟡 50% (2/4) 🟡 50% (2/4)
loki 🔴 0% (0/1) 🔴 0% (0/1) 🔴 0% (0/1) 🔴 0% (0/1) 🔴 0% (0/1) 🔴 0% (0/1) 🔴 0% (0/1) 🔴 0% (0/1)
medium 🟡 83% (⅚) 🟡 67% (4/6) 🟡 83% (⅚) 🟡 67% (4/6) 🟡 50% (3/6) 🟡 83% (⅚) 🟡 67% (4/6) 🟡 67% (4/6)
metrics 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1)
network 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🔴 0% (0/1) 🟢 100% (1/1)
one-test 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🔴 0% (0/1) 🟢 100% (1/1)
port-forward 🟡 50% (½) 🟡 50% (½) 🟡 50% (½) 🟡 50% (½) 🟡 50% (½) 🟡 50% (½) 🟡 50% (½) 🟡 50% (½)
question-answer 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1) 🟢 100% (1/1)
regression 🟡 67% (6/9) 🟡 78% (7/9) 🟡 78% (7/9) 🟡 89% (8/9) 🟡 89% (8/9) 🟡 67% (6/9) 🟡 33% (3/9) 🟡 67% (6/9)
runbooks 🟢 100% (2/2) 🟡 50% (½) 🟢 100% (2/2) 🟡 50% (½) 🟢 100% (2/2) 🟡 50% (½) 🟡 50% (½) 🟡 50% (½)
Overall 🟡 71% (10/14) 🟡 71% (10/14) 🟡 79% (11/14) 🟡 79% (11/14) 🟡 71% (10/14) 🟡 71% (10/14) 🟡 50% (7/14) 🟡 64% (9/14)

Raw Results

Status of all evaluations across models. Color coding:

  • 🟢 Passing 100% (stable)
  • 🟡 Passing 1-99%
  • 🔴 Passing 0% (failing)
  • 🔧 Mock data failure (missing or invalid test data)
  • ⚠️ Setup failure (environment/infrastructure issue)
  • ⏱️ Timeout or rate limit error
  • ⏭️ Test skipped (e.g., known issue or precondition not met)
Eval ID deepseek/deepseek-chat deepseek/deepseek-reasoner eu.anthropic.claude-haiku-4-5-20251001-v1:0 eu.anthropic.claude-opus-4-5-20251101-v1:0 eu.anthropic.claude-sonnet-4-5-20250929-v1:0 gemini/gemini-3-flash-preview gemini/gemini-3-pro-preview gpt-5.2
09_crashpod 🔗 🟢 🟢 🟢 🟢 🟢 🟢 🔴 🟢
101_loki_historical_logs_pod_deleted 🔗 🔴 🔴 🔴 🔴 🔴 🔴 🔴 🔴
108_logs_nearby_lines 🔗 🔴 🔴 🔴 🔴 🔴 🟢 🟢 🔴
111_pod_names_contain_service 🔗 🟢 🟢 🟢 🟢 🟢 🟢 🟢 🟢
12_job_crashing 🔗 🔴 🟢 🟢 🟢 🟢 🔴 🟢 🔴
162_get_runbooks 🔗 🟢 🟢 🟢 🟢 🟢 🟢 🔴 🟢
176_network_policy_blocking_traffic_no_runbooks 🔗 🟢 🟢 🟢 🟢 🟢 🟢 🔴 🟢
179_grafana_big_dashboard_query 🔗 🟢 🟢 🟢 🟢 🟢 🟢 🟢 🟢
24_misconfigured_pvc 🔗 🟢 🔴 🔴 🟢 🟢 🟢 🟢 🟢
43_current_datetime_from_prompt 🔗 🔴 🟢 🟢 🟢 🟢 🔴 🔴 🔴
61_exact_match_counting 🔗 🟢 🟢 🟢 🟢 🟢 🟢 🔴 🟢
73a_time_window_anomaly 🔗 🟢 🟢 🟢 🟢 🔴 🟢 🔴 🟢
73b_time_window_anomaly 🔗 🟢 🟢 🟢 🟢 🔴 🟢 🟢 🟢
96_no_matching_runbook 🔗 🟢 🔴 🟢 🔴 🟢 🔴 🟢 🔴
SUMMARY 🟡 71% (10/14) 🟡 71% (10/14) 🟡 79% (11/14) 🟡 79% (11/14) 🟡 71% (10/14) 🟡 71% (10/14) 🟡 50% (7/14) 🟡 64% (9/14)

Detailed Raw Results

Eval ID deepseek/deepseek-chat deepseek/deepseek-reasoner eu.anthropic.claude-haiku-4-5-20251001-v1:0 eu.anthropic.claude-opus-4-5-20251101-v1:0 eu.anthropic.claude-sonnet-4-5-20250929-v1:0 gemini/gemini-3-flash-preview gemini/gemini-3-pro-preview gpt-5.2
09_crashpod 🔗 🟢 100% (1/1) / ⏱️ 100.4s / 💰 $0.01 🟢 100% (1/1) / ⏱️ 197.2s / 💰 $0.01 🟢 100% (1/1) / ⏱️ 26.1s / 💰 $0.05 🟢 100% (1/1) / ⏱️ 29.8s / 💰 $0.22 🟢 100% (1/1) / ⏱️ 27.9s / 💰 $0.15 🟢 100% (1/1) / ⏱️ 68.4s / 💰 $0.04 🔴 0% (0/1) / ⏱️ 284.7s 🟢 100% (1/1) / ⏱️ 50.5s / 💰 $0.11
101_loki_historical_logs_pod_deleted 🔗 🔴 0% (0/1) / ⏱️ 156.6s / 💰 $0.02 🔴 0% (0/1) / ⏱️ 361.8s / 💰 $0.03 🔴 0% (0/1) / ⏱️ 28.7s / 💰 $0.05 🔴 0% (0/1) / ⏱️ 194.7s / 💰 $0.40 🔴 0% (0/1) / ⏱️ 40.1s / 💰 $0.18 🔴 0% (0/1) / ⏱️ 258.0s 🔴 0% (0/1) / ⏱️ 340.5s / 💰 $0.31 🔴 0% (0/1) / ⏱️ 39.4s / 💰 $0.14
108_logs_nearby_lines 🔗 🔴 0% (0/1) / ⏱️ 210.1s / 💰 $0.04 🔴 0% (0/1) / ⏱️ 567.0s / 💰 $0.04 🔴 0% (0/1) / ⏱️ 33.6s / 💰 $0.07 🔴 0% (0/1) / ⏱️ 44.0s / 💰 $0.28 🔴 0% (0/1) / ⏱️ 44.2s / 💰 $0.27 🟢 100% (1/1) / ⏱️ 72.5s / 💰 $0.09 🟢 100% (1/1) / ⏱️ 912.4s / 💰 $0.28 🔴 0% (0/1) / ⏱️ 38.9s / 💰 $0.09
111_pod_names_contain_service 🔗 🟢 100% (1/1) / ⏱️ 140.8s / 💰 $0.02 🟢 100% (1/1) / ⏱️ 260.8s / 💰 $0.01 🟢 100% (1/1) / ⏱️ 28.0s / 💰 $0.06 🟢 100% (1/1) / ⏱️ 27.5s / 💰 $0.19 🟢 100% (1/1) / ⏱️ 34.4s / 💰 $0.18 🟢 100% (1/1) / ⏱️ 51.8s / 💰 $0.04 🟢 100% (1/1) / ⏱️ 1415.8s / 💰 $0.12 🟢 100% (1/1) / ⏱️ 41.2s / 💰 $0.09
12_job_crashing 🔗 🔴 0% (0/1) / ⏱️ 119.8s / 💰 $0.01 🟢 100% (1/1) / ⏱️ 317.3s / 💰 $0.02 🟢 100% (1/1) / ⏱️ 33.0s / 💰 $0.06 🟢 100% (1/1) / ⏱️ 39.0s / 💰 $0.26 🟢 100% (1/1) / ⏱️ 38.7s / 💰 $0.19 🔴 0% (0/1) / ⏱️ 452.8s 🟢 100% (1/1) / ⏱️ 1523.9s / 💰 $0.23 🔴 0% (0/1) / ⏱️ 33.9s / 💰 $0.07
162_get_runbooks 🔗 🟢 100% (1/1) / ⏱️ 119.2s / 💰 $0.02 🟢 100% (1/1) / ⏱️ 253.2s / 💰 $0.03 🟢 100% (1/1) / ⏱️ 33.2s / 💰 $0.08 🟢 100% (1/1) / ⏱️ 50.3s / 💰 $0.36 🟢 100% (1/1) / ⏱️ 43.2s / 💰 $0.22 🟢 100% (1/1) / ⏱️ 54.1s / 💰 $0.04 🔴 0% (0/1) / ⏱️ 626.3s / 💰 $0.21 🟢 100% (1/1) / ⏱️ 52.6s / 💰 $0.16
176_network_policy_blocking_traffic_no_runbooks 🔗 🟢 100% (1/1) / ⏱️ 129.5s / 💰 $0.01 🟢 100% (1/1) / ⏱️ 255.1s / 💰 $0.01 🟢 100% (1/1) / ⏱️ 38.3s / 💰 $0.07 🟢 100% (1/1) / ⏱️ 50.4s / 💰 $0.32 🟢 100% (1/1) / ⏱️ 42.9s / 💰 $0.22 🟢 100% (1/1) / ⏱️ 71.6s / 💰 $0.05 🔴 0% (0/1) / ⏱️ 10.4s / 💰 $0.03 🟢 100% (1/1) / ⏱️ 34.7s / 💰 $0.08
179_grafana_big_dashboard_query 🔗 🟢 100% (1/1) / ⏱️ 63.5s / 💰 $0.01 🟢 100% (1/1) / ⏱️ 138.4s / 💰 $0.01 🟢 100% (1/1) / ⏱️ 26.5s / 💰 $0.05 🟢 100% (1/1) / ⏱️ 29.9s / 💰 $0.20 🟢 100% (1/1) / ⏱️ 22.0s / 💰 $0.12 🟢 100% (1/1) / ⏱️ 75.8s / 💰 $0.03 🟢 100% (1/1) / ⏱️ 911.5s / 💰 $0.13 🟢 100% (1/1) / ⏱️ 25.1s / 💰 $0.09
24_misconfigured_pvc 🔗 🟢 100% (1/1) / ⏱️ 102.6s / 💰 $0.01 🔴 0% (0/1) / ⏱️ 33.9s / 💰 $0.00 🔴 0% (0/1) / ⏱️ 4.9s / 💰 $0.02 🟢 100% (1/1) / ⏱️ 33.0s / 💰 $0.22 🟢 100% (1/1) / ⏱️ 34.9s / 💰 $0.17 🟢 100% (1/1) / ⏱️ 35.5s / 💰 $0.04 🟢 100% (1/1) / ⏱️ 1110.0s / 💰 $0.13 🟢 100% (1/1) / ⏱️ 36.0s / 💰 $0.07
43_current_datetime_from_prompt 🔗 🔴 0% (0/1) / ⏱️ 19.4s / 💰 $0.00 🟢 100% (1/1) / ⏱️ 7.8s / 💰 $0.00 🟢 100% (1/1) / ⏱️ 6.5s / 💰 $0.02 🟢 100% (1/1) / ⏱️ 6.0s / 💰 $0.10 🟢 100% (1/1) / ⏱️ 5.0s / 💰 $0.07 🔴 0% (0/1) / ⏱️ 19.6s / 💰 $0.03 🔴 0% (0/1) / ⏱️ 44.1s / 💰 $0.05 🔴 0% (0/1) / ⏱️ 12.0s / 💰 $0.02
61_exact_match_counting 🔗 🟢 100% (1/1) / ⏱️ 70.1s / 💰 $0.01 🟢 100% (1/1) / ⏱️ 86.6s / 💰 $0.01 🟢 100% (1/1) / ⏱️ 10.8s / 💰 $0.03 🟢 100% (1/1) / ⏱️ 17.2s / 💰 $0.16 🟢 100% (1/1) / ⏱️ 22.2s / 💰 $0.09 🟢 100% (1/1) / ⏱️ 31.1s / 💰 $0.02 🔴 0% (0/1) / ⏱️ 284.4s 🟢 100% (1/1) / ⏱️ 16.6s / 💰 $0.07
73a_time_window_anomaly 🔗 🟢 100% (1/1) / ⏱️ 174.4s / 💰 $0.02 🟢 100% (1/1) / ⏱️ 201.2s / 💰 $0.01 🟢 100% (1/1) / ⏱️ 27.6s / 💰 $0.05 🟢 100% (1/1) / ⏱️ 33.8s / 💰 $0.22 🔴 0% (0/1) / ⏱️ 33.4s / 💰 $0.27 🟢 100% (1/1) / ⏱️ 57.4s / 💰 $0.06 🔴 0% (0/1) / ⏱️ 1303.9s 🟢 100% (1/1) / ⏱️ 51.4s / 💰 $0.17
73b_time_window_anomaly 🔗 🟢 100% (1/1) / ⏱️ 153.1s / 💰 $0.02 🟢 100% (1/1) / ⏱️ 246.0s / 💰 $0.02 🟢 100% (1/1) / ⏱️ 24.9s / 💰 $0.05 🟢 100% (1/1) / ⏱️ 40.0s / 💰 $0.24 🔴 0% (0/1) / ⏱️ 36.0s / 💰 $0.29 🟢 100% (1/1) / ⏱️ 36.7s / 💰 $0.06 🟢 100% (1/1) / ⏱️ 180.9s / 💰 $0.27 🟢 100% (1/1) / ⏱️ 44.5s / 💰 $0.11
96_no_matching_runbook 🔗 🟢 100% (1/1) / ⏱️ 160.5s / 💰 $0.03 🔴 0% (0/1) / ⏱️ 366.0s / 💰 $0.03 🟢 100% (1/1) / ⏱️ 52.2s / 💰 $0.11 🔴 0% (0/1) / ⏱️ 83.8s / 💰 $0.48 🟢 100% (1/1) / ⏱️ 57.7s / 💰 $0.29 🔴 0% (0/1) / ⏱️ 105.7s / 💰 $0.12 🟢 100% (1/1) / ⏱️ 883.4s / 💰 $0.29 🔴 0% (0/1) / ⏱️ 73.8s / 💰 $0.24

Results are automatically generated and updated weekly. View full traces and detailed analysis in Braintrust experiment: ci-benchmark-21401358384.