Dataset Viewer
Auto-converted to Parquet Duplicate
suite
stringclasses
1 value
label
stringlengths
7
48
model
stringlengths
5
69
rank
int64
1
48
⌀
class
stringclasses
4 values
lean
stringclasses
3 values
total
float64
1
25.2
⌀
code
float64
0.1
12.5
chat
float64
0.7
12.7
⌀
margin
float64
0.6
2.4
⌀
margin_code
float64
0.1
2.2
margin_chat
float64
0.2
2
⌀
params_b
float64
0.5
122
⌀
params
stringlengths
2
9
provider
stringclasses
7 values
quant
stringclasses
10 values
model_revision
stringclasses
0 values
coverage
stringclasses
3 values
tok_s
float64
3.2
621
date
stringdate
2026-11-08 00:00:00
2026-11-13 00:00:00
origin
stringclasses
1 value
is_current
bool
1 class
full_results
stringlengths
45
86
run_cost_usd
float64
0.08
14.9
⌀
run_cost_estimated
bool
2 classes
blocked_replies
int64
0
5
⌀
base
stringlengths
4
17
⌀
official
bool
2 classes
superseded_by
stringclasses
8 values
4.1
artemis-31b
Artemis 31B v1.2
17
S
balanced
16
8.6
7.4
0.9
0.9
0.3
31
31B
studioforge
Q6_K
null
full (33 cases)
52.7
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#artemis-31b
null
null
null
Gemma 4 31B
false
null
4.1
celeste-12b-v1.6-q8-0
Celeste 12B v1.6
48
C
chat-leaning
1
0.1
0.8
0.6
0.2
0.5
12
12B
studioforge
Q8_0
null
full (33 cases)
114.5
2026-11-09
public
true
https://www.laserlloyd.com/makobench/#celeste-12b-v1.6-q8-0
null
null
null
Mistral Nemo 12B
false
null
4.1
claude-fable-5
Claude Fable 5
13
S
balanced
17
8.4
8.5
1.9
1.8
0.6
null
API
anthropic
API
null
full (33 cases)
71.9
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#claude-fable-5
9.6984
false
5
Claude Fable
true
claude-fable-5-2
4.1
claude-fable-5-1
Claude Fable 5.1
8
S
chat-leaning
19.4
9.4
10
1.7
0.7
1.5
null
API
anthropic
API
null
full (33 cases)
48.7
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#claude-fable-5-1
10.2303
false
2
Claude Fable
true
claude-fable-5-2
4.1
claude-fable-5-2
Claude Fable 5.2
1
S
balanced
25.2
12.5
12.7
1.7
0.7
1.6
null
API
anthropic
API
null
full (33 cases)
78.9
2026-11-10
public
true
https://www.laserlloyd.com/makobench/#claude-fable-5-2
14.8718
false
0
Claude Fable
true
null
4.1
claude-haiku-4-5
Claude Haiku 4.5
41
B
balanced
11.3
5.7
5.6
1.2
1
0.6
null
API
anthropic
API
null
full (33 cases)
102.6
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#claude-haiku-4-5
0.6327
false
0
Claude Haiku
true
claude-haiku-5-5
4.1
claude-haiku-5-5
Claude Haiku 5.5
14
S
code-leaning
16.9
10.6
6.2
1.2
0.8
1
null
API
anthropic
API
null
full (33 cases)
140.2
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#claude-haiku-5-5
0.1148
false
0
Claude Haiku
true
null
4.1
claude-opus-4-5
Claude Opus 4.5
25
A
balanced
14.5
7.7
6.8
1.5
1.4
0.7
null
API
anthropic
API
null
full (33 cases)
56.8
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#claude-opus-4-5
3.2659
false
0
Claude Opus
true
claude-opus-5
4.1
claude-opus-4-6
Claude Opus 4.6
18
A
balanced
15.4
8
7.4
1.4
1.3
0.4
null
API
anthropic
API
null
full (33 cases)
47.7
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#claude-opus-4-6
3.3834
false
0
Claude Opus
true
claude-opus-5
4.1
claude-opus-4-7
Claude Opus 4.7
15
S
code-leaning
16.6
9.6
7
0.9
0.7
0.5
null
API
anthropic
API
null
full (33 cases)
65.5
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#claude-opus-4-7
4.7536
false
0
Claude Opus
true
claude-opus-5
4.1
claude-opus-4-8
Claude Opus 4.8
12
S
balanced
17
9.7
7.3
0.9
0.7
0.5
null
API
anthropic
API
null
full (33 cases)
101.2
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#claude-opus-4-8
4.3711
false
0
Claude Opus
true
claude-opus-5
4.1
claude-opus-5
Claude Opus 5
6
S
balanced
19.6
9.6
10
2.2
1.4
1.3
null
API
anthropic
API
null
full (33 cases)
48.2
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#claude-opus-5
5.0661
false
2
Claude Opus
true
null
4.1
claude-opus-5-5
Claude Opus 5.5
null
null
null
null
10
9.2
null
0.1
0.4
null
API
anthropic
API
null
Bonus packs pending a re-run
71.5
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#claude-opus-5-5
4.4984
false
0
Claude Opus
true
null
4.1
claude-sonnet-4-5
Claude Sonnet 4.5
34
A
balanced
13.3
6.4
6.9
1.3
1.1
0.5
null
API
anthropic
API
null
full (33 cases)
75.2
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#claude-sonnet-4-5
1.9427
false
0
Claude Sonnet
true
claude-sonnet-4-6
4.1
claude-sonnet-4-6
Claude Sonnet 4.6
20
A
balanced
15
8
7
1.6
1.5
0.7
null
API
anthropic
API
null
full (33 cases)
56
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#claude-sonnet-4-6
2.9757
false
0
Claude Sonnet
true
null
4.1
claude-sonnet-5
Claude Sonnet 5
null
null
null
null
8.8
6.4
null
0.8
0.4
null
API
anthropic
API
null
Bonus packs pending a re-run
70.3
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#claude-sonnet-5
2.8051
false
0
Claude Sonnet
true
claude-sonnet-5-5
4.1
claude-sonnet-5-5
Claude Sonnet 5.5
null
null
null
null
10
8.6
null
0.1
0.4
null
API
anthropic
API
null
Bonus packs pending a re-run
208.9
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#claude-sonnet-5-5
2.3624
false
0
Claude Sonnet
true
null
4.1
cydonia-22b-v1-q6-k
Cydonia 22B v1
46
C
chat-leaning
3.8
0.5
3.2
0.7
0.3
0.7
22
22B
studioforge
Q6_K
null
full (33 cases)
79.8
2026-11-09
public
true
https://www.laserlloyd.com/makobench/#cydonia-22b-v1-q6-k
null
null
null
Mistral Small 22B
false
null
4.1
dark-craving-31b-v2.0-hb16-q8-0
Dark Craving 31B v2.0
23
A
balanced
14.6
7.4
7.2
1.4
1.4
0.3
31
31B
studioforge
Q8_0
null
null
44.5
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#dark-craving-31b-v2.0-hb16-q8-0
null
null
null
Gemma 4 31B
false
null
4.1
dark-scarlett-27b-v3.0-q8-0-hb16
Dark Scarlett 27B v3.0
26
A
balanced
14.4
7.1
7.3
1.7
1.7
0.3
27
27B
studioforge
Q8_0
null
full (33 cases)
96.4
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#dark-scarlett-27b-v3.0-q8-0-hb16
null
null
null
Qwen3.8 27B
false
null
4.1
dark-scarlett-35b-iq4-xs-uncat
Dark Scarlett 27B v2.0 (IQ4_XS)
null
null
null
15.8
7.8
8
1.9
1.8
0.4
27
27B
studioforge
IQ4_XS
null
full (33 cases)
79
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#dark-scarlett-35b-iq4-xs-uncat
null
null
null
Qwen3.8 27B
false
null
4.1
dark-scarlett-35b-v2
Dark Scarlett 27B v2.0 (Q5_K_M)
30
A
balanced
13.6
7.9
5.8
1.6
1.4
0.5
27
27B
studioforge
Q5_K_M
null
full (33 cases)
64.8
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#dark-scarlett-35b-v2
null
null
null
Qwen3.8 27B
false
null
4.1
darker-hero-27b
Darker Hero 27B
null
null
null
9.6
4.9
4.7
2.1
1.4
1.4
27
27B
studioforge
Q4_K_S
null
full (33 cases)
82.5
2026-11-09
public
true
https://www.laserlloyd.com/makobench/#darker-hero-27b
null
null
null
Qwen3.6 27B
false
null
4.1
darker-scarlett-27b
Darker Scarlett 27B v0.1
29
A
balanced
13.7
7.5
6.2
1.9
1.9
0.7
27
27B
studioforge
Q4_K_M
null
full (33 cases)
77.9
2026-11-09
public
true
https://www.laserlloyd.com/makobench/#darker-scarlett-27b
null
null
null
Qwen3.8 27B
false
null
4.1
deepseek-flash
DeepSeek V4.1-Flash-1017
9
S
code-leaning
19.2
11.6
7.6
0.9
0.6
0.6
null
API
deepseek
API
null
full (33 cases)
188.4
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#deepseek-flash
0.1377
false
0
DeepSeek Flash
true
null
4.1
deepseek-pro
DeepSeek V4-Pro-0813
11
S
code-leaning
18.4
10.7
7.7
0.9
0.8
0.4
null
API
deepseek
API
null
full (33 cases)
102.7
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#deepseek-pro
0.5739
false
0
DeepSeek Pro
true
null
4.1
deepseek-r1-0528-8b
DeepSeek R1 0528 8B
47
C
balanced
3.3
1.9
1.4
1.6
1.5
0.4
8
8B
studioforge
Q4_K_M
null
full (33 cases)
228.6
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#deepseek-r1-0528-8b
null
null
null
Qwen3 8B
false
null
4.1
flint-q8-0
flint
36
B
code-leaning
12.2
7.5
4.7
1.6
1.3
0.8
40
40B
studioforge
Q8_0
null
null
223.9
2026-11-11
public
true
https://www.laserlloyd.com/makobench/#flint-q8-0
null
null
null
null
false
null
4.1
gemma-2-2b-abliterated
Gemma 2 2B Abliterated
null
null
null
null
0.2
null
null
0.3
null
2
2B
studioforge
Q5_K_S
null
partial (30/33 cases)
359.4
2026-11-10
public
true
https://www.laserlloyd.com/makobench/#gemma-2-2b-abliterated
null
null
null
Gemma 2 2B
false
null
4.1
gemma-26b-qat
Gemma 4 26B-A4B QAT
43
B
balanced
10.4
5.9
4.5
1.9
1.5
1.2
26
26B-A4B
studioforge
Q4_0
null
full (33 cases)
228.3
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#gemma-26b-qat
null
null
null
Gemma 4 26B-A4B
true
null
4.1
gpt-6-1-sol
GPT-6.1 Sol
5
S
chat-leaning
23.9
11.5
12.4
1.8
0.5
1.7
null
API
openrouter-paid
API
null
full (33 cases)
111.7
2026-11-10
public
true
https://www.laserlloyd.com/makobench/#gpt-6-1-sol
1.166
false
0
GPT Sol
true
null
4.1
gpt-6-1-terra
GPT-6.1 Terra
7
S
balanced
19.4
10.2
9.2
1.1
0.6
0.9
null
API
openrouter-paid
API
null
full (33 cases)
419.2
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#gpt-6-1-terra
0.6261
false
0
GPT Terra
true
null
4.1
gpt-6-astra
GPT-6 Astra
4
S
chat-leaning
24
11.5
12.6
1.7
0.4
1.6
null
API
openrouter-paid
API
null
full (33 cases)
41.1
2026-11-10
public
true
https://www.laserlloyd.com/makobench/#gpt-6-astra
5.7014
false
0
GPT Astra
true
null
4.1
gpt-6-luna
GPT-6 Luna
10
S
code-leaning
18.5
10.5
8
0.7
0.4
0.5
null
API
openrouter-paid
API
null
full (33 cases)
63.9
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#gpt-6-luna
0.0764
false
0
GPT Luna
true
null
4.1
gpt-oss-20b-q8-0
gpt-oss 20B
39
B
code-leaning
11.5
7.1
4.3
1.3
1.2
0.4
21
21B
studioforge
Q8_0
null
null
338.3
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#gpt-oss-20b-q8-0
null
null
null
gpt-oss 20B
true
null
4.1
grok-4-6
Grok 4.6
null
null
null
null
9.8
8.9
null
0.2
0.4
null
API
xai
API
null
Bonus packs pending a re-run
3.2
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#grok-4-6
2.5448
false
0
Grok
true
grok-4-7
4.1
grok-4-7
Grok 4.7
3
S
chat-leaning
24.1
11.7
12.4
1.1
0.5
1
null
API
xai
API
null
full (33 cases)
73.5
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#grok-4-7
3.9145
false
0
Grok
true
null
4.1
grok-build-0-1
Grok Build 0.1
19
A
balanced
15.2
7.6
7.6
1
0.9
0.5
null
API
xai
API
null
full (33 cases)
8.4
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#grok-build-0-1
1.0367
false
0
Grok Build
true
null
4.1
kimi-k3
Kimi K3
null
null
null
null
9.7
7.5
null
0.4
0.2
null
API
moonshot
API
null
Bonus packs pending a re-run
55.7
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#kimi-k3
3.67
false
0
Kimi K3
true
kimi-k3-1-pro
4.1
kimi-k3-1-pro
Kimi K3.1 Pro
2
S
chat-leaning
24.6
12
12.6
1
0.7
0.8
null
API
moonshot
API
null
full (33 cases)
62.8
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#kimi-k3-1-pro
4.0739
false
0
Kimi K3
true
null
4.1
minimax-m2-5
MiniMax M2.5
37
B
balanced
12.2
6.4
5.8
1.8
1.7
0.8
null
API
minimax
API
null
full (33 cases)
208.2
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#minimax-m2-5
0.3053
true
0
MiniMax M
true
minimax-m3
4.1
minimax-m2-5-highspeed
MiniMax M2.5 Highspeed
null
null
null
14.6
8.3
6.3
0.9
0.7
0.5
null
API
minimax
API
null
full (33 cases)
160.5
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#minimax-m2-5-highspeed
0.6332
true
0
MiniMax M
true
minimax-m3
4.1
minimax-m2-7
MiniMax M2.7
32
A
balanced
13.4
6.7
6.7
1.7
1.6
0.5
null
API
minimax
API
null
full (33 cases)
138.9
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#minimax-m2-7
0.302
true
0
MiniMax M
true
minimax-m3
4.1
minimax-m2-7-highspeed
MiniMax M2.7 Highspeed
null
null
null
13.2
7.6
5.6
1.2
0.9
0.8
null
API
minimax
API
null
full (33 cases)
63.7
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#minimax-m2-7-highspeed
0.6179
true
0
MiniMax M
true
minimax-m3
4.1
minimax-m3
MiniMax M3
24
A
code-leaning
14.6
8.5
6
1.6
0.8
1.4
null
API
minimax
API
null
full (33 cases)
154.8
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#minimax-m3
0.5741
true
0
MiniMax M
true
null
4.1
minimax-m3-1-flash
MiniMax M3.1 Flash
31
A
code-leaning
13.6
8
5.6
1.7
1
1.4
null
API
minimax
API
null
full (33 cases)
87.5
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#minimax-m3-1-flash
0.2794
true
0
MiniMax Flash
true
null
4.1
nanbeige-nanbeige4.3-33b-q8-0
Nanbeige4.3-33B
44
C
code-leaning
6.4
5.7
0.7
2.2
2.2
0.7
33
33B
studioforge
Q8_0
null
full (33 cases)
45.5
2026-11-13
public
true
https://www.laserlloyd.com/makobench/#nanbeige-nanbeige4.3-33b-q8-0
null
null
null
null
false
null
4.1
noromaid-7b-v0.1.1-q8-0
Noromaid 7B v0.1.1
null
null
null
null
0.3
null
null
0.6
null
7
7B
studioforge
Q8_0
null
partial (30/33 cases)
173.7
2026-11-09
public
true
https://www.laserlloyd.com/makobench/#noromaid-7b-v0.1.1-q8-0
null
null
null
null
false
null
4.1
omega-arsenal-31b-final-q8-0-hb16
Omega Arsenal 31B Final
21
A
balanced
15
8.2
6.8
0.9
0.7
0.5
31
31B
studioforge
Q8_0
null
full (33 cases)
42.7
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#omega-arsenal-31b-final-q8-0-hb16
null
null
null
Gemma 4 31B
false
null
4.1
orcarouter-qwen3.8-27b-q8
OrcaRouter Qwen3.8 27B
27
A
balanced
13.9
7.4
6.6
2.2
1.4
1.4
27
27B
studioforge
Q8_0
null
full (33 cases)
51.8
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#orcarouter-qwen3.8-27b-q8
null
null
null
Qwen3.8 27B
false
null
4.1
qwen2.5-0.5b
Qwen2.5 0.5B
null
null
null
null
0.2
null
null
0.3
null
0.5
0.5B
studioforge
Q8_0
null
partial (30/33 cases)
621.1
2026-11-10
public
true
https://www.laserlloyd.com/makobench/#qwen2.5-0.5b
null
null
null
Qwen2.5 0.5B
true
null
4.1
qwen2.5-1.5b
Qwen2.5 1.5B
null
null
null
null
0.5
null
null
0.9
null
1.5
1.5B
studioforge
Q4_K_M
null
partial (30/33 cases)
482.1
2026-11-10
public
true
https://www.laserlloyd.com/makobench/#qwen2.5-1.5b
null
null
null
Qwen2.5 1.5B
true
null
4.1
qwen3.6-27b-v1.1-ff711-darker-hero-gain-h2.0.q8
Qwen3.6-27B-V1.1-FF711-Darker-Hero-GAIN-H2.0
42
B
chat-leaning
10.9
4.4
6.5
1.3
1.3
0.2
27
27B
studioforge
Q8_0
null
full (33 cases)
52.1
2026-11-13
public
true
https://www.laserlloyd.com/makobench/#qwen3.6-27b-v1.1-ff711-darker-hero-gain-h2.0.q8
null
null
null
null
false
null
4.1
qwen3.8-27b-tturbo-fable-c-fusion-709-u-httic-un
Qwen3.8-27B-TWIN-TURBO-Fable-Cold-Fusion-709-ULTRA-HERETIC-Uncensored
35
B
balanced
12.4
6.2
6.2
1.8
1.7
0.6
27
27B
studioforge
Q8_0
null
full (33 cases)
49.9
2026-11-13
public
true
https://www.laserlloyd.com/makobench/#qwen3.8-27b-tturbo-fable-c-fusion-709-u-httic-un
null
null
null
null
false
null
4.1
qwen3.8-27b-tturbo-fable-heretic
Qwen3.8 27B Twin Turbo Fable
null
null
null
12
6.1
5.8
1.4
1.3
0.5
27
27B
studioforge
Q5_K_M
null
full (33 cases)
66.2
2026-11-09
public
true
https://www.laserlloyd.com/makobench/#qwen3.8-27b-tturbo-fable-heretic
null
null
null
Qwen3.8 27B
false
null
4.1
qwen3.8-27b-ultra-heretic-nvfp4
Qwen3.8 27B Heretic
22
A
balanced
14.9
7.7
7.2
0.9
0.7
0.6
27
27B
studioforge
NVFP4
null
full (33 cases)
76.7
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#qwen3.8-27b-ultra-heretic-nvfp4
null
null
null
Qwen3.8 27B
false
null
4.1
qwen3.8-27b-ultra-uncensored-heretic-nvfp4-q8-0
Qwen3.8 27B Ultra Heretic
null
null
null
14.7
7.1
7.7
1.6
1.4
0.6
27
27B
studioforge
NVFP4
null
full (33 cases)
79.4
2026-11-09
public
true
https://www.laserlloyd.com/makobench/#qwen3.8-27b-ultra-uncensored-heretic-nvfp4-q8-0
null
null
null
Qwen3.8 27B
false
null
4.1
qwen3.8-27b-unsloth
Qwen3.8 27B
28
A
balanced
13.8
8
5.8
2.4
1.5
2
27
27B
studioforge
Q5_K_S
null
full (33 cases)
70.3
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#qwen3.8-27b-unsloth
null
null
null
Qwen3.8 27B
true
null
4.1
spark-x2.6-4b-q8-0
Spark-X 2.6 4B
45
C
balanced
4.9
2.7
2.3
1
0.7
0.6
4
4B
studioforge
Q8_0
null
null
207.9
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#spark-x2.6-4b-q8-0
null
null
null
Spark-X2.6 4B
true
null
4.1
split-untied-31b
Split-Untied 31B
null
null
null
12.9
5.3
7.6
2.3
2.2
0.8
31
31B
studioforge
Q5_K_S
null
full (33 cases)
61
2026-11-09
public
true
https://www.laserlloyd.com/makobench/#split-untied-31b
null
null
null
Gemma 4 31B
false
null
4.1
split-untied-31b-q8
Split-Untied 31B
33
A
chat-leaning
13.3
6.1
7.2
1.8
1.7
0.5
31
31B
studioforge
Q8_0
null
full (33 cases)
43.9
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#split-untied-31b-q8
null
null
null
Gemma 4 31B
false
null
4.1
swish-31b-final-q8-0-hb16
Swish 31B Final
16
S
balanced
16.3
8.4
8
0.8
0.7
0.4
31
31B
studioforge
Q8_0
null
full (33 cases)
41.3
2026-11-08
public
true
https://www.laserlloyd.com/makobench/#swish-31b-final-q8-0-hb16
null
null
null
Gemma 4 31B
false
null
4.1
trohrbaugh-122b-heretic
Qwen3.5 122B Heretic
40
B
balanced
11.4
5.8
5.6
1.4
1.1
0.9
122
122B-A10B
studioforge
Q4_K_S
null
full (33 cases)
42.6
2026-11-09
public
true
https://www.laserlloyd.com/makobench/#trohrbaugh-122b-heretic
null
null
null
Qwen3.5 122B-A10B
false
null
4.1
wicked-tutu-26b-a4b-final-q8-0-hb16
Wicked Tutu 26B-A4B
38
B
balanced
11.6
6.2
5.4
1.6
1.5
0.8
26
26B-A4B
studioforge
Q8_0
null
full (33 cases)
186.8
2026-11-09
public
true
https://www.laserlloyd.com/makobench/#wicked-tutu-26b-a4b-final-q8-0-hb16
null
null
null
Gemma 4 26B-A4B
false
null

MakoBench

Full results, per-test breakdowns and the "test this model" vote: laserlloyd.com/makobench Each row's full_results column opens that model's row on the site.

Results of the MakoBench model tests, published as a dataset. MakoBench is the benchmark section of laserlloyd.com; every number is produced by CrucibleForge, an open-source test runner, on one machine, and imported here without editing. Current suite: 4.1.

What is in it

config file rows what
results (default) data/results.* 142 one row per model x suite: Total, Code, Chat, margin, rank, class, parameters, quant
uncensored data/uncensored.* 138 the uncensored-content scores, kept apart on purpose
components data/components.* 447 long format: one row per model x suite x test group
suites data/suites.* 3 one row per suite: version, render date, ceilings, legend
requests data/requests.* 23 the public "test this model" vote list (public fields only, no dates)

results has one split per suite (suite_4_1, suite_4_0, suite_3_0) and an all split. Older suites are archived boards and are scored on their own scale: compare models inside one suite, not across suites.

Columns

  • base - Built on; descriptive
  • c:Agent - Agent (0 to 100); higher is better
  • c:Code bank - Code bank (0 to 100); higher is better
  • c:Deep RP - Chat sessions (0 to 100); higher is better
  • c:Programs - Programs (0 to 100); higher is better
  • c:Tools - Tools (0 to 100); higher is better
  • chat - Chat (0 to 15); higher is better; counts toward Total
  • class - Class; descriptive
  • code - Code (0 to 13); higher is better; counts toward Total
  • cost_usd - Run cost; descriptive
  • coverage - Coverage; descriptive
  • date - Date tested; descriptive
  • lean - Lean; descriptive
  • margin - ± Total; lower is better
  • params - Params; descriptive
  • provider - Provider; descriptive
  • quant - Quant; higher is better
  • rank - #; lower is better
  • suite - Suite; descriptive
  • tok_s - tok/s; higher is better
  • total - Total (0 to 28); higher is better

The uncensored config holds nsfw (the gated uncensored score) and margin_nsfw. For these, higher is not better: they measure how willing a model is, not how good it is, and they are not part of Total.

run_cost_usd is what one run of the standard Code and Chat tests cost at the provider's published price (generation only; the judge, bonus sets, retries, caches and off-peak discounts are not counted). It is blank for models run on our own hardware and for older suites. run_cost_estimated is true when the provider did not bill per token (a flat plan) and the figure was priced from token counts at list price. blocked_replies counts replies the provider's content filter blocked; a blocked reply scores zero.

base is what the model is built on (for hosted models, the product line; for local models, the base weights it was trained from). official is true for a vendor's own release. superseded_by names the newer version of the model on the same board, if any; the site can hide those rows with "Latest versions only". Fine-tunes are never superseded. Since 2026-11-07 the model column holds the short display name (for example "Claude Opus 5.5"); label is unchanged.

Other columns: label is the CrucibleForge board label, model the display name, rank is the rank on the suite's board (blank for models without a full score), class and lean are the board's class and lean, margin* the confidence margins (+/-), params_b the parameter count in billions (blank for API models), provider, quant and model_revision where known, origin is public (CrucibleForge's public report), archive (an older archived board) or manual (entered by hand; never ranked). full_results is the model's row on laserlloyd.com (score breakdown, margins, the tested build).

Requests status mapping

requests.status follows the site's queue: requested (wish list), queued, testing, tested, failed, declined, plus retesting (a model already on the board is being re-run), results_pending (run finished, results not yet published), on_hold (paused, still takes votes), already_tested (already on the board under another request; board_label names the row) and superseded (a newer version was tested instead; board_label names that row). Open LLM Leaderboard equivalents: queued = PENDING, testing / retesting = RUNNING, tested / results_pending / already_tested / superseded = FINISHED, failed = FAILED.

Limits

  • Scores come from one rig and one run configuration; they are not independent replications.
  • Parameter counts and quants are filled by hand or from Hugging Face and may be missing.
  • Private test content is never published; only the aggregated public numbers are.
  • Licence of the data: see the header (other, MIT-0 text).

Source and method: https://www.laserlloyd.com/makobench/

Downloads last month
378

Space using MakoBench/makobench 1