Muse Ember (xhigh) leads on average score with 8.8 vs 8.7. Muse Ember (xhigh) has the lower benchmark cost at $0.232 vs ~$0.298. Muse Ember (xhigh) is faster at 13.90s vs 166.34s, with pass rates of 76.8% vs 78.3%.
Last updated at:
2026-11-10
Compared models
Rank
#73
Total Output Tokens
409,112
Response Time (avg)
13.90s
Total Cost
$0.232
Candidate responses were generated on local hardware. OpenRouter was used only for judging. qwen3.8:latest · Q4_K_M · Ollama 0.32.12 · NVIDIA GeForce RTX 3090.
Rank
#81
Total Output Tokens
677,080
Response Time (avg)
166.34s
Total Cost
~$0.298Estimated GPU electricity cost only. It excludes the rest of the computer and hardware ownership. NVIDIA GeForce RTX 3090: 3.0102 h × 0.300 kW × €0.2896/kWh × 1.138 USD/EUR = ~$0.297613.
Recommended modelMuse Ember (xhigh)
It has the best score here (8.8), while responding about 12.0x faster than Qwen3.8 27B (high).
Qwen3.8 27BQwen3.8 27BhighCandidate responses were generated on local hardware. OpenRouter was used only for judging. qwen3.8:latest · Q4_K_M · Ollama 0.32.12 · NVIDIA GeForce RTX 3090.Release: 2026-08-14Free Available
Qwen3.8 27BQwen3.8 27BhighCandidate responses were generated on local hardware. OpenRouter was used only for judging. qwen3.8:latest · Q4_K_M · Ollama 0.32.12 · NVIDIA GeForce RTX 3090.Release: 2026-08-14Free Available
Score
8.8Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
8.7Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
Rank
#73
#81
Reliability
10.0First-attempt success score: 10.0 means no retryable target API or rate-limit failures before successful calls; tracked failures lower the score.…
9.7First-attempt success score: 10.0 means no retryable target API or rate-limit failures before successful calls; tracked failures lower the score.…
Consistency
8.9Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
9.2Consistency score reflects run-to-run stability (10 = very consistent, even if consistently wrong).…
Attempts
69/69
69/69
Tests Correct
A test is fully passed only if every run passed for that test.Wrong answer: 5Did not follow instructions: 1No answer: 1Response Time (avg)13.90sResponse Time (max)79.38sResponse Time (total)319.61sA test is fully passed only if every run passed for that test.…
A test is fully passed only if every run passed for that test.Did not follow instructions: 2No answer: 2Wrong answer: 2Response Time (avg)166.34sResponse Time (max)640.85sResponse Time (total)3825.81sA test is fully passed only if every run passed for that test.…
Attempt pass rate
76.8%Attempt pass rate = passed attempts / total attempts across runs.…
78.3%Attempt pass rate = passed attempts / total attempts across runs.…
Flaky tests
3Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
2Flaky tests had mixed outcomes across runs (at least one pass and one fail).…
Total Runs
69Total Runs…
69Total Runs…
Cost per result
1.450
~1.751Estimated GPU electricity cost only. It excludes the rest of the computer and hardware ownership. NVIDIA GeForce RTX 3090: 3.0102 h × 0.300 kW × €0.2896/kWh × 1.138 USD/EUR = ~$0.297613.
Total Cost
$0.232
~$0.298Estimated GPU electricity cost only. It excludes the rest of the computer and hardware ownership. NVIDIA GeForce RTX 3090: 3.0102 h × 0.300 kW × €0.2896/kWh × 1.138 USD/EUR = ~$0.297613.
Input Price
$0.100 / 1MInput Price…
N/AInput Price…
Output Price
$0.500 / 1MOutput Price…
N/AOutput Price…
Cache Read Price
$0.004 / 1MCache Read Price…
N/ACache Read Price…
Cache Write Price
$0.125 / 1MCache Write Price…
N/ACache Write Price…
Total Input Tokens
274,208Total Input Tokens…
348,967Total Input Tokens…
Output Tokens
11,740Output Tokens…
1,005Output Tokens…
Reasoning Tokens
397,372Reasoning Tokens…
676,075Reasoning Tokens…
Response Time (avg)
13.90sResponse Time (avg)…
166.34sResponse Time (avg)…
Response Time (max)
79.38sResponse Time (max)…
640.85sResponse Time (max)…
Response Time (total)
319.61sResponse Time (total)…
3825.81sResponse Time (total)…
Parameters
~150B total (~10B active)
27.3B
Availability
Closed
Weights available
Cache prices apply to input tokens. Reads reuse cached prompts; writes store them and can cost extra. Output tokens use the output price.
Model generation showcase
Hamster playing table tennis
Prompt: Create a detailed SVG illustration of a hamster playing table tennis.
#73 Muse Ember
xhigh
Cost
$0.008
Time
24.3s
Tokens
14,705 tok
#81 Qwen3.8 27B
highCandidate responses were generated on local hardware. OpenRouter was used only for judging. qwen3.8:latest · Q4_K_M · Ollama 0.32.12 · NVIDIA GeForce RTX 3090.
Cost
~$0.008Estimated GPU electricity cost only. It excludes the rest of the computer and hardware ownership. NVIDIA GeForce RTX 3090: 0.0750 h × 0.300 kW × €0.2896/kWh × 1.138 USD/EUR = ~$0.007419.
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)47.42sResponse Time (max)47.42sResponse Time (total)47.42sA test is fully passed only if every run passed for that test.…
47.42sResponse Time (avg)…
151,581Total Input Tokens…
2,661Output Tokens…
2,824Reasoning Tokens…
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging. qwen3.8:latest · Q4_K_M · Ollama 0.32.12 · NVIDIA GeForce RTX 3090.
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)132.27sResponse Time (max)132.27sResponse Time (total)132.27sA 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)2.05sResponse Time (max)3.33sResponse Time (total)8.19sA test is fully passed only if every run passed for that test.…
2.05sResponse Time (avg)…
798Total Input Tokens…
288Output Tokens…
5,861Reasoning Tokens…
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging. qwen3.8:latest · Q4_K_M · Ollama 0.32.12 · NVIDIA GeForce RTX 3090.
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)8.10sResponse Time (max)21.48sResponse Time (total)32.40sA 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)10.41sResponse Time (max)13.59sResponse Time (total)31.23sA test is fully passed only if every run passed for that test.…
10.41sResponse Time (avg)…
7,410Total Input Tokens…
1,225Output Tokens…
40,702Reasoning Tokens…
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging. qwen3.8:latest · Q4_K_M · Ollama 0.32.12 · NVIDIA GeForce RTX 3090.
8.1Summarizes broad quality across our full private benchmark suite, so ranking reflects consistent performance.…
7.0Consistency 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.Did not follow instructions: 1Response Time (avg)248.62sResponse Time (max)458.25sResponse Time (total)745.85sA 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)13.24sResponse Time (max)20.41sResponse Time (total)26.47sA test is fully passed only if every run passed for that test.…
13.24sResponse Time (avg)…
91,473Total Input Tokens…
4,513Output Tokens…
21,164Reasoning Tokens…
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging. qwen3.8:latest · Q4_K_M · Ollama 0.32.12 · NVIDIA GeForce RTX 3090.
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)185.57sResponse Time (max)333.21sResponse Time (total)371.15sA 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)3.59sResponse Time (max)5.45sResponse Time (total)7.18sA test is fully passed only if every run passed for that test.…
3.59sResponse Time (avg)…
7,215Total Input Tokens…
368Output Tokens…
5,817Reasoning Tokens…
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging. qwen3.8:latest · Q4_K_M · Ollama 0.32.12 · NVIDIA GeForce RTX 3090.
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)34.85sResponse Time (max)57.71sResponse Time (total)69.71sA test is fully passed only if every run passed for that test.…
5.3Summarizes 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).…
44.4%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: 2Response Time (avg)23.77sResponse Time (max)37.24sResponse Time (total)71.31sA test is fully passed only if every run passed for that test.…
23.77sResponse Time (avg)…
873Total Input Tokens…
1,044Output Tokens…
76,708Reasoning Tokens…
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging. qwen3.8:latest · Q4_K_M · Ollama 0.32.12 · NVIDIA GeForce RTX 3090.
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.No answer: 1Wrong answer: 1Response Time (avg)526.74sResponse Time (max)640.85sResponse Time (total)1580.21sA 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)2.88sResponse Time (max)2.88sResponse Time (total)2.88sA test is fully passed only if every run passed for that test.…
2.88sResponse Time (avg)…
522Total Input Tokens…
214Output Tokens…
2,610Reasoning Tokens…
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging. qwen3.8:latest · Q4_K_M · Ollama 0.32.12 · NVIDIA GeForce RTX 3090.
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)8.45sResponse Time (max)8.45sResponse Time (total)8.45sA test is fully passed only if every run passed for that test.…
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.Did not follow instructions: 1Response Time (avg)4.17sResponse Time (max)5.79sResponse Time (total)8.34sA test is fully passed only if every run passed for that test.…
4.17sResponse Time (avg)…
759Total Input Tokens…
241Output Tokens…
7,345Reasoning Tokens…
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging. qwen3.8:latest · Q4_K_M · Ollama 0.32.12 · NVIDIA GeForce RTX 3090.
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)8.24sResponse Time (max)10.18sResponse Time (total)16.49sA test is fully passed only if every run passed for that test.…
5.9Summarizes 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).…
55.6%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.No answer: 1Wrong answer: 1Response Time (avg)29.02sResponse Time (max)79.38sResponse Time (total)87.07sA test is fully passed only if every run passed for that test.…
29.02sResponse Time (avg)…
786Total Input Tokens…
504Output Tokens…
202,040Reasoning Tokens…
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging. qwen3.8:latest · Q4_K_M · Ollama 0.32.12 · NVIDIA GeForce RTX 3090.
7.6Summarizes 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).…
77.8%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.No answer: 1Response Time (avg)199.90sResponse Time (max)569.33sResponse Time (total)599.69sA 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)3.93sResponse Time (max)3.93sResponse Time (total)3.93sA test is fully passed only if every run passed for that test.…
3.93sResponse Time (avg)…
12,563Total Input Tokens…
448Output Tokens…
1,643Reasoning Tokens…
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging. qwen3.8:latest · Q4_K_M · Ollama 0.32.12 · NVIDIA GeForce RTX 3090.
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)5.62sResponse Time (max)5.62sResponse Time (total)5.62sA 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.Wrong answer: 1Response Time (avg)25.61sResponse Time (max)25.61sResponse Time (total)25.61sA test is fully passed only if every run passed for that test.…
25.61sResponse Time (avg)…
228Total Input Tokens…
234Output Tokens…
30,658Reasoning Tokens…
Qwen3.8 27BCandidate responses were generated on local hardware. OpenRouter was used only for judging. qwen3.8:latest · Q4_K_M · Ollama 0.32.12 · NVIDIA GeForce RTX 3090.
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.Wrong answer: 1Response Time (avg)263.98sResponse Time (max)263.98sResponse Time (total)263.98sA test is fully passed only if every run passed for that test.…