The average score is effectively tied at 7.2 vs 7.2. Qwen3.8 27B (medium) has the lower benchmark cost at ~$0.073 vs $0.449. Step 3.7 Flash (low) is faster at 20.72s vs 40.09s, with pass rates of 66.7% vs 63.8%.
Last updated at:
2026-10-01
Compared models
Candidate responses were generated on local hardware. OpenRouter was used only for judging.
Rank
#141
Total Output Tokens
170,696
Response Time (avg)
40.09s
Total Cost
~$0.073Estimated GPU electricity cost only. It excludes the rest of the computer and hardware ownership. NVIDIA GeForce RTX 3090: 0.7349 h × 0.300 kW × €0.2896/kWh × 1.138 USD/EUR = ~$0.072660.
Rank
#147
Total Output Tokens
339,108
Response Time (avg)
20.72s
Total Cost
$0.449
Recommended modelQwen3.8 27B (medium)
It has the best score here (7.2), while costing about 6.2x less than Step 3.7 Flash (low).
Detailed comparison
Metric
Qwen3.8 27BQwen3.8 27BmediumCandidate responses were generated on local hardware. OpenRouter was used only for judging.Release: 2026-08-14Free Available
Qwen3.8 27BQwen3.8 27BmediumCandidate responses were generated on local hardware. OpenRouter was used only for judging.Release: 2026-08-14Free Available
A test is fully passed only if every run passed for that test.Wrong answer: 9Invalid tool call: 2No answer: 1Response Time (avg)20.72sResponse Time (max)124.75sResponse Time (total)476.67sA test is fully passed only if every run passed for that test.…
Attempt pass rate
66.7%Attempt pass rate = passed attempts / total attempts across runs.…
63.8%Attempt pass rate = passed attempts / total attempts across runs.…
Flaky tests
2Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
6Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
Total Runs
69Total Runs…
69Total Runs…
Cost per result
~0.520Estimated GPU electricity cost only. It excludes the rest of the computer and hardware ownership. NVIDIA GeForce RTX 3090: 0.7349 h × 0.300 kW × €0.2896/kWh × 1.138 USD/EUR = ~$0.072660.
4.078
Total Cost
~$0.073Estimated GPU electricity cost only. It excludes the rest of the computer and hardware ownership. NVIDIA GeForce RTX 3090: 0.7349 h × 0.300 kW × €0.2896/kWh × 1.138 USD/EUR = ~$0.072660.
$0.449
Input Price
N/AInput Price…
$0.200 / 1MInput Price…
Output Price
N/AOutput Price…
$1.150 / 1MOutput Price…
Total Input Tokens
277,164Total Input Tokens…
292,685Total Input Tokens…
Output Tokens
1,913Output Tokens…
339,108Output Tokens…
Reasoning Tokens
168,783Reasoning Tokens…
0Reasoning Tokens…
Response Time (avg)
40.09sResponse Time (avg)…
20.72sResponse Time (avg)…
Response Time (max)
214.63sResponse Time (max)…
124.75sResponse Time (max)…
Response Time (total)
881.89sResponse Time (total)…
476.67sResponse Time (total)…
Parameters
27.3B
198B total (11B active)
Availability
Weights available
Open source
Model generation showcase
Hamster playing table tennis
Prompt: Create a detailed SVG illustration of a hamster playing table tennis.
#141 Qwen3.8 27B
mediumCandidate responses were generated on local hardware. OpenRouter was used only for judging.
Cost
~$0.002Estimated GPU electricity cost only. It excludes the rest of the computer and hardware ownership. NVIDIA GeForce RTX 3090: 0.0121 h × 0.300 kW × €0.2896/kWh × 1.138 USD/EUR = ~$0.001193.
Time
43.4s
Tokens
2,956 tok
#147 Step 3.7 Flash
low
Invalid SVG
Cost
$0.004
Time
25.3s
Tokens
3,072 tok
Rank
-
Cost
-
Time
-
Tokens
-
Top Models by Score
Score vs Total Cost
Response Time (avg)
Score vs Response Time (avg)
Total Output Tokens
Score vs Total Output Tokens
Category Breakdown
Agentic
Score
Consistency
Attempt pass rate
Flaky tests
Tests Correct
Response Time (avg)
Input Tokens
Output Tokens
Reasoning Tokens
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging.
4.7Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
1.6Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
66.7%Attempt pass rate = passed attempts / total attempts across runs.…
1Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.Wrong answer: 1Response Time (avg)187.85sResponse Time (max)187.85sResponse Time (total)187.85sA test is fully passed only if every run passed for that test.…
7.4Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
3.8Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
66.7%Attempt pass rate = passed attempts / total attempts across runs.…
1Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.Invalid tool call: 1Response Time (avg)83.63sResponse Time (max)83.63sResponse Time (total)83.63sA test is fully passed only if every run passed for that test.…
83.63sResponse Time (avg)…
188,843Total Input Tokens…
18,715Output Tokens…
0Reasoning Tokens…
Anti-AI Tricks
Score
Consistency
Attempt pass rate
Flaky tests
Tests Correct
Response Time (avg)
Input Tokens
Output Tokens
Reasoning Tokens
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging.
10.0Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
10.0Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
100.0%Attempt pass rate = passed attempts / total attempts across runs.…
0Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.No failed answers.Response Time (avg)2.93sResponse Time (max)5.18sResponse Time (total)11.70sA test is fully passed only if every run passed for that test.…
8.7Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
7.9Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
91.7%Attempt pass rate = passed attempts / total attempts across runs.…
1Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.Wrong answer: 1Response Time (avg)4.02sResponse Time (max)12.52sResponse Time (total)16.10sA test is fully passed only if every run passed for that test.…
4.02sResponse Time (avg)…
756Total Input Tokens…
10,896Output Tokens…
0Reasoning Tokens…
Coding
Score
Consistency
Attempt pass rate
Flaky tests
Tests Correct
Response Time (avg)
Input Tokens
Output Tokens
Reasoning Tokens
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging.
10.0Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
10.0Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
100.0%Attempt pass rate = passed attempts / total attempts across runs.…
0Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.No failed answers.Response Time (avg)40.50sResponse Time (max)72.14sResponse Time (total)121.51sA test is fully passed only if every run passed for that test.…
8.2Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
7.2Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
88.9%Attempt pass rate = passed attempts / total attempts across runs.…
1Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.Wrong answer: 1Response Time (avg)9.46sResponse Time (max)12.69sResponse Time (total)28.38sA test is fully passed only if every run passed for that test.…
9.46sResponse Time (avg)…
7,437Total Input Tokens…
18,685Output Tokens…
0Reasoning Tokens…
Combined
Score
Consistency
Attempt pass rate
Flaky tests
Tests Correct
Response Time (avg)
Input Tokens
Output Tokens
Reasoning Tokens
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging.
8.7Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
6.9Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
83.3%Attempt pass rate = passed attempts / total attempts across runs.…
1Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.Invalid tool call: 1Response Time (avg)124.48sResponse Time (max)214.63sResponse Time (total)248.96sA test is fully passed only if every run passed for that test.…
7.3Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
5.8Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
83.3%Attempt pass rate = passed attempts / total attempts across runs.…
1Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.Invalid tool call: 1Response Time (avg)66.18sResponse Time (max)124.39sResponse Time (total)132.37sA test is fully passed only if every run passed for that test.…
66.18sResponse Time (avg)…
77,415Total Input Tokens…
93,682Output Tokens…
0Reasoning Tokens…
Data parsing and extraction
Score
Consistency
Attempt pass rate
Flaky tests
Tests Correct
Response Time (avg)
Input Tokens
Output Tokens
Reasoning Tokens
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging.
6.5Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
10.0Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
50.0%Attempt pass rate = passed attempts / total attempts across runs.…
0Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.Wrong answer: 1Response Time (avg)8.76sResponse Time (max)10.36sResponse Time (total)17.53sA test is fully passed only if every run passed for that test.…
7.3Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
5.8Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
83.3%Attempt pass rate = passed attempts / total attempts across runs.…
1Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.Wrong answer: 1Response Time (avg)2.29sResponse Time (max)3.15sResponse Time (total)4.58sA test is fully passed only if every run passed for that test.…
2.29sResponse Time (avg)…
7,398Total Input Tokens…
2,667Output Tokens…
0Reasoning Tokens…
Domain specific
Score
Consistency
Attempt pass rate
Flaky tests
Tests Correct
Response Time (avg)
Input Tokens
Output Tokens
Reasoning Tokens
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging.
5.3Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
10.0Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
33.3%Attempt pass rate = passed attempts / total attempts across runs.…
0Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.Wrong answer: 2Response Time (avg)76.98sResponse Time (max)144.07sResponse Time (total)230.93sA test is fully passed only if every run passed for that test.…
2.9Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
7.2Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
11.1%Attempt pass rate = passed attempts / total attempts across runs.…
1Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.Wrong answer: 3Response Time (avg)22.65sResponse Time (max)45.52sResponse Time (total)67.95sA test is fully passed only if every run passed for that test.…
22.65sResponse Time (avg)…
837Total Input Tokens…
48,299Output Tokens…
0Reasoning Tokens…
General Intelligence
Score
Consistency
Attempt pass rate
Flaky tests
Tests Correct
Response Time (avg)
Input Tokens
Output Tokens
Reasoning Tokens
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging.
5.0Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
10.0Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
0.0%Attempt pass rate = passed attempts / total attempts across runs.…
0Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.Did not follow instructions: 1Response Time (avg)9.20sResponse Time (max)9.20sResponse Time (total)9.20sA test is fully passed only if every run passed for that test.…
3.4Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
9.3Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
0.0%Attempt pass rate = passed attempts / total attempts across runs.…
0Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.Wrong answer: 1Response Time (avg)7.00sResponse Time (max)7.00sResponse Time (total)7.00sA test is fully passed only if every run passed for that test.…
7.00sResponse Time (avg)…
525Total Input Tokens…
4,604Output Tokens…
0Reasoning Tokens…
Instructions following
Score
Consistency
Attempt pass rate
Flaky tests
Tests Correct
Response Time (avg)
Input Tokens
Output Tokens
Reasoning Tokens
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging.
10.0Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
10.0Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
100.0%Attempt pass rate = passed attempts / total attempts across runs.…
0Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.No failed answers.Response Time (avg)2.54sResponse Time (max)2.67sResponse Time (total)5.07sA test is fully passed only if every run passed for that test.…
9.8Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
10.0Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
100.0%Attempt pass rate = passed attempts / total attempts across runs.…
0Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.No failed answers.Response Time (avg)1.58sResponse Time (max)1.80sResponse Time (total)3.16sA test is fully passed only if every run passed for that test.…
1.58sResponse Time (avg)…
735Total Input Tokens…
1,857Output Tokens…
0Reasoning Tokens…
Puzzle Solving
Score
Consistency
Attempt pass rate
Flaky tests
Tests Correct
Response Time (avg)
Input Tokens
Output Tokens
Reasoning Tokens
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging.
8.3Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
10.0Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
66.7%Attempt pass rate = passed attempts / total attempts across runs.…
0Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.Did not follow instructions: 1Response Time (avg)12.62sResponse Time (max)29.62sResponse Time (total)37.85sA test is fully passed only if every run passed for that test.…
5.5Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
9.9Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
33.3%Attempt pass rate = passed attempts / total attempts across runs.…
0Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.Wrong answer: 2Response Time (avg)1.84sResponse Time (max)3.42sResponse Time (total)5.52sA test is fully passed only if every run passed for that test.…
1.84sResponse Time (avg)…
756Total Input Tokens…
3,564Output Tokens…
0Reasoning Tokens…
Tool Calling
Score
Consistency
Attempt pass rate
Flaky tests
Tests Correct
Response Time (avg)
Input Tokens
Output Tokens
Reasoning Tokens
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging.
3.0Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
10.0Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
0.0%Attempt pass rate = passed attempts / total attempts across runs.…
0Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.API error: 1Response Time (avg)0msResponse Time (max)0msResponse Time (total)0msA test is fully passed only if every run passed for that test.…
10.0Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
10.0Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
100.0%Attempt pass rate = passed attempts / total attempts across runs.…
0Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.No failed answers.Response Time (avg)3.25sResponse Time (max)3.25sResponse Time (total)3.25sA test is fully passed only if every run passed for that test.…
3.25sResponse Time (avg)…
7,746Total Input Tokens…
1,360Output Tokens…
0Reasoning Tokens…
Trivia
Score
Consistency
Attempt pass rate
Flaky tests
Tests Correct
Response Time (avg)
Input Tokens
Output Tokens
Reasoning Tokens
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging.
3.0Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
10.0Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
0.0%Attempt pass rate = passed attempts / total attempts across runs.…
0Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.No answer: 1Response Time (avg)11.29sResponse Time (max)11.29sResponse Time (total)11.29sA test is fully passed only if every run passed for that test.…
3.0Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
10.0Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
0.0%Attempt pass rate = passed attempts / total attempts across runs.…
0Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
A test is fully passed only if every run passed for that test.No answer: 1Response Time (avg)124.75sResponse Time (max)124.75sResponse Time (total)124.75sA test is fully passed only if every run passed for that test.…