MLX Model Test Report

Generated 2026-02-27 09:26 · AFM MLX Backend · v0.9.5-4d58e3f

mlx-model-test.sh --prompts Scripts/test-Qwen3.5-35B-A3B-4bit.txt --smart 1:claude,codex

Test Runs
184
Passed
184
Failed
0
Best tok/s
145.3
Fastest
mlx-community/Qwen3.5-35B-A3B-4bit @ stop-long-phrase

Performance Ranking (by tokens/sec)

Click a row to jump to its full response below.

# Model / Config Status codexclaude Temp Load (s) Tokens Gen (s) Tokens/sec Prompt
1 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-long-phrase stop-long-phrase OK 23 0.0 1.0 1143 7.87
145.3
Write a 3-paragraph essay about renewable energy. Start the ...
2 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-double-newline stop-double-newline OK 23 0.0 1.0 742 5.8
128.0
Write a short paragraph about the ocean. Then write a second...
3 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-immediate stop-immediate OK 23 0.0 1.0 767 6.11
125.6
Explain what gravity is.
4 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-newline stop-newline OK 23 0.0 1.0 817 6.55
124.8
Name three primary colors and explain why they matter in des...
5 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-period stop-period OK 43 0.0 1.0 692 5.55
124.7
Tell me about the sun in three sentences.
6 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-period stop-period OK 23 0.0 1.0 831 6.69
124.3
Name three primary colors and explain why they matter in des...
7 mlx-community/Qwen3.5-35B-A3B-4bit @ guided-json-simple guided-json-simple
--guided-json '{"type":"object","properties":{"name":{"type":"string"},"age":{"type":"integer"}},"required":["name","age"]}'
OK 23 0.0 1.0 1373 11.06
124.1
Generate a person record.
8 mlx-community/Qwen3.5-35B-A3B-4bit @ seed-42-run2 seed-42-run2
--seed 42
OK 43 0.7 1.0 4096 33.15
123.6
Write a limerick about a cat.
9 mlx-community/Qwen3.5-35B-A3B-4bit @ no-penalty no-penalty
--presence-penalty 0.0
OK 33 0.8 1.0 3795 30.75
123.4
Write a long essay about the history of bread making across ...
10 mlx-community/Qwen3.5-35B-A3B-4bit @ scientist scientist OK 43 0.3 1.0 2402 19.47
123.4
Tell me about quantum computing.
11 mlx-community/Qwen3.5-35B-A3B-4bit @ guided-json-nested guided-json-nested
--guided-json '{"type":"object","properties":{"city":{"type":"string"},"population":{"type":"integer"},"landmarks":{"type":"array","items":{"type":"string"}}},"required":["city","population","landmarks"]}'
OK 23 0.0 1.0 937 7.6
123.3
Describe Tokyo as a structured record with at least 3 landma...
12 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-double-newline stop-double-newline OK 23 0.0 1.0 817 6.62
123.3
Name three primary colors and explain why they matter in des...
13 mlx-community/Qwen3.5-35B-A3B-4bit @ response-format-json response-format-json OK 23 0.0 1.0 1303 10.57
123.3
Return a JSON object with keys "language", "year_created", a...
14 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-four-max stop-four-max OK 23 0.0 1.0 4096 33.25
123.2
List 10 facts about numbers, alternating between digit forma...
15 mlx-community/Qwen3.5-35B-A3B-4bit @ kv-quantized kv-quantized
--kv-bits 4
OK 43 0.7 2.0 1132 9.2
123.1
Summarize the key ideas of machine learning in a few paragra...
16 mlx-community/Qwen3.5-35B-A3B-4bit @ developer-role developer-role OK 33 0.0 1.0 4096 33.28
123.1
developer: You are a helpful coding assistant. Only respond ...
17 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-immediate stop-immediate OK 23 0.0 1.0 817 6.64
123.0
Name three primary colors and explain why they matter in des...
18 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-four-max stop-four-max OK 23 0.0 1.0 818 6.65
123.0
Name three primary colors and explain why they matter in des...
19 mlx-community/Qwen3.5-35B-A3B-4bit @ eli5 eli5 OK 23 0.7 1.0 921 7.5
122.8
Tell me about quantum computing.
20 mlx-community/Qwen3.5-35B-A3B-4bit @ small-kv small-kv
--max-kv-size 2048
OK 43 0.7 1.0 1184 9.65
122.7
Summarize the key ideas of machine learning in a few paragra...
21 mlx-community/Qwen3.5-35B-A3B-4bit @ json-output json-output OK 23 0.0 1.0 384 3.13
122.6
Respond with a valid JSON object containing keys "name", "ag...
22 mlx-community/Qwen3.5-35B-A3B-4bit @ think-raw think-raw
--raw
OK 33 0.0 1.0 833 6.8
122.5
Solve step by step: If a train travels at 60 mph for 2.5 hou...
23 mlx-community/Qwen3.5-35B-A3B-4bit @ response-format-schema response-format-schema OK 33 0.0 1.0 1478 12.07
122.5
Generate a fictional person profile.
24 mlx-community/Qwen3.5-35B-A3B-4bit @ seed-42-run1 seed-42-run1
--seed 42
OK 23 0.7 1.0 4096 33.45
122.5
Write a limerick about a cat.
25 mlx-community/Qwen3.5-35B-A3B-4bit @ non-streaming-seeded non-streaming-seeded
--no-streaming --seed 123
OK 33 0.7 1.0 2469 20.17
122.4
Write a 4-line poem about the ocean.
26 mlx-community/Qwen3.5-35B-A3B-4bit @ eli5 eli5 OK 53 0.7 1.0 4096 33.48
122.3
Name three primary colors and explain why they matter in des...
27 mlx-community/Qwen3.5-35B-A3B-4bit @ agent-cached-turn3 agent-cached-turn3
--enable-prefix-caching
OK 43 0.0 1.0 356 2.92
122.1
Write a unit test for the timeout feature you just described...
28 mlx-community/Qwen3.5-35B-A3B-4bit @ prefill-small-256 prefill-small-256
--prefill-step-size 256
OK 23 0.0 1.0 3652 29.91
122.1
Given this architecture, what are the top 3 performance bott...
29 mlx-community/Qwen3.5-35B-A3B-4bit @ max-completion-tokens max-completion-tokens OK 33 0.0 1.0 2300 18.87
121.9
Explain the causes and consequences of the French Revolution...
30 mlx-community/Qwen3.5-35B-A3B-4bit @ max-completion-tokens max-completion-tokens OK 33 0.0 1.0 1174 9.65
121.7
max_completion_tokens: 100
31 mlx-community/Qwen3.5-35B-A3B-4bit @ no-streaming no-streaming
--no-streaming
OK 23 0.7 1.0 4096 33.69
121.6
Write a short poem about the moon.
32 mlx-community/Qwen3.5-35B-A3B-4bit @ greedy greedy OK 23 0.0 1.0 546 4.5
121.3
Explain why the sky is blue in exactly 3 bullet points.
33 mlx-community/Qwen3.5-35B-A3B-4bit @ prefill-large-4096 prefill-large-4096
--prefill-step-size 4096
OK 23 0.0 1.0 3652 30.1
121.3
Given this architecture, what are the top 3 performance bott...
34 mlx-community/Qwen3.5-35B-A3B-4bit @ min-p min-p
--min-p 0.05
OK 43 0.8 1.0 793 6.54
121.3
Explain why the sky is blue in exactly 3 bullet points.
35 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-guided-json-comma stop-guided-json-comma
--guided-json '{"type":"object","properties":{"name":{"type":"string"},"age":{"type":"integer"},"city":{"type":"string"}},"required":["name","age","city"]}'
OK 53 0.0 1.0 826 6.81
121.3
Name three primary colors and explain why they matter in des...
36 mlx-community/Qwen3.5-35B-A3B-4bit @ streaming-seeded streaming-seeded
--seed 123
OK 33 0.7 1.0 2469 20.36
121.2
Write a 4-line poem about the ocean.
37 mlx-community/Qwen3.5-35B-A3B-4bit @ raw-mode raw-mode
--raw
OK 43 0.7 1.0 4096 33.78
121.2
Think step by step: what is 17 * 23?
38 mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-xml tool-call-xml
--tool-call-parser xmlFunction
OK 33 0.0 1.0 4096 33.82
121.1
What is the weather in Paris in celsius?
39 mlx-community/Qwen3.5-35B-A3B-4bit @ math math OK 33 0.0 2.0 1830 15.12
121.0
A store sells notebooks for $3.50 each. If you buy 5 or more...
40 mlx-community/Qwen3.5-35B-A3B-4bit @ pirate pirate OK 23 0.7 1.0 1459 12.06
121.0
Tell me about quantum computing.
41 mlx-community/Qwen3.5-35B-A3B-4bit @ think-normal think-normal OK 33 0.0 1.0 833 6.89
121.0
Solve step by step: If a train travels at 60 mph for 2.5 hou...
42 mlx-community/Qwen3.5-35B-A3B-4bit @ pirate pirate OK 53 0.7 1.0 1836 15.18
121.0
Name three primary colors and explain why they matter in des...
43 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-json-object-key stop-json-object-key OK 23 0.0 1.0 4096 33.89
120.9
Generate a JSON object with keys "name", "age", and "city" f...
44 mlx-community/Qwen3.5-35B-A3B-4bit @ very-verbose very-verbose
--very-verbose
OK 33 0.7 1.0 658 5.46
120.5
Hello, how are you?
45 mlx-community/Qwen3.5-35B-A3B-4bit @ numbered-list numbered-list OK 23 0.0 1.0 489 4.06
120.4
List exactly 5 animals, one per line, numbered 1-5. No other...
46 mlx-community/Qwen3.5-35B-A3B-4bit @ prefill-small-256 prefill-small-256
--prefill-step-size 256
OK 43 0.0 1.0 2610 21.7
120.3
Name three primary colors and explain why they matter in des...
47 mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-none tool-call-none OK 33 0.0 1.0 475 3.95
120.2
What is the current temperature in Berlin?
48 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-system-pirate stop-system-pirate OK 23 0.0 1.0 4096 34.09
120.2
Name three primary colors and explain why they matter in des...
49 mlx-community/Qwen3.5-35B-A3B-4bit @ no-penalty no-penalty
--presence-penalty 0.0
OK 23 0.8 1.0 1580 13.16
120.0
Name three primary colors and explain why they matter in des...
50 mlx-community/Qwen3.5-35B-A3B-4bit @ high-temp high-temp OK 53 1.5 1.0 1357 11.31
119.9
Name three primary colors and explain why they matter in des...
51 mlx-community/Qwen3.5-35B-A3B-4bit @ prefill-default prefill-default OK 23 0.0 1.0 3652 30.46
119.9
Given this architecture, what are the top 3 performance bott...
52 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-non-streaming stop-non-streaming
--no-streaming
OK 23 0.0 1.0 4096 34.2
119.8
List 10 planets or celestial objects, numbered 1 through 10,...
53 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-guided-json-brace stop-guided-json-brace
--guided-json '{"type":"object","properties":{"color":{"type":"string"},"hex":{"type":"string"}},"required":["color","hex"]}'
OK 23 0.0 1.0 1063 8.87
119.8
Describe the color blue with its hex code.
54 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-high-temp stop-high-temp OK 23 1.0 1.0 1771 14.8
119.6
Name three primary colors and explain why they matter in des...
55 mlx-community/Qwen3.5-35B-A3B-4bit @ response-format-text response-format-text OK 33 0.0 1.0 370 3.1
119.5
Tell me the language Python, what year it was created, and w...
56 mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-complex tool-call-complex OK 33 0.0 1.0 458 3.84
119.4
Search recursively for all Swift files under Sources/ with a...
57 mlx-community/Qwen3.5-35B-A3B-4bit @ scientist scientist OK 53 0.3 1.0 2950 24.72
119.3
Name three primary colors and explain why they matter in des...
58 mlx-community/Qwen3.5-35B-A3B-4bit @ high-temp high-temp OK 23 1.5 1.0 852 7.15
119.1
Explain why the sky is blue in exactly 3 bullet points.
59 mlx-community/Qwen3.5-35B-A3B-4bit @ no-streaming no-streaming
--no-streaming
OK 43 0.7 1.0 1623 13.63
119.1
Name three primary colors and explain why they matter in des...
60 mlx-community/Qwen3.5-35B-A3B-4bit @ seed-42-run1 seed-42-run1
--seed 42
OK 53 0.7 1.0 1135 9.55
118.8
Name three primary colors and explain why they matter in des...
61 mlx-community/Qwen3.5-35B-A3B-4bit @ seed-42-run2 seed-42-run2
--seed 42
OK 23 0.7 1.0 1135 9.55
118.8
Name three primary colors and explain why they matter in des...
62 mlx-community/Qwen3.5-35B-A3B-4bit @ streaming-seeded streaming-seeded
--seed 123
OK 33 0.7 1.0 1501 12.67
118.5
Name three primary colors and explain why they matter in des...
63 mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-auto tool-call-auto OK 33 0.0 1.0 502 4.24
118.5
What is the weather in Tokyo?
64 mlx-community/Qwen3.5-35B-A3B-4bit @ verbose verbose
--verbose
OK 33 0.7 1.0 387 3.27
118.4
Hello, how are you?
65 mlx-community/Qwen3.5-35B-A3B-4bit @ very-verbose very-verbose
--very-verbose
OK 33 0.7 1.0 1894 16.0
118.4
Name three primary colors and explain why they matter in des...
66 mlx-community/Qwen3.5-35B-A3B-4bit @ default default OK 43 0.7 1.0 1090 9.21
118.4
Explain why the sky is blue in exactly 3 bullet points.
67 mlx-community/Qwen3.5-35B-A3B-4bit @ prefill-large-4096 prefill-large-4096
--prefill-step-size 4096
OK 53 0.0 1.0 2610 22.05
118.3
Name three primary colors and explain why they matter in des...
68 mlx-community/Qwen3.5-35B-A3B-4bit @ minimal-prompt minimal-prompt OK 33 0.7 1.0 390 3.3
118.2
Hi
69 mlx-community/Qwen3.5-35B-A3B-4bit @ non-streaming-seeded non-streaming-seeded
--no-streaming --seed 123
OK 33 0.7 1.0 1501 12.72
118.0
Name three primary colors and explain why they matter in des...
70 mlx-community/Qwen3.5-35B-A3B-4bit @ top-k top-k
--top-k 30
OK 43 0.8 2.0 625 5.3
117.9
Explain why the sky is blue in exactly 3 bullet points.
71 mlx-community/Qwen3.5-35B-A3B-4bit @ guided-json-nested guided-json-nested
--guided-json '{"type":"object","properties":{"city":{"type":"string"},"population":{"type":"integer"},"landmarks":{"type":"array","items":{"type":"string"}}},"required":["city","population","landmarks"]}'
OK 23 0.0 1.0 810 6.88
117.7
Name three primary colors and explain why they matter in des...
72 mlx-community/Qwen3.5-35B-A3B-4bit @ developer-role developer-role OK 33 0.0 1.0 437 3.72
117.6
Write a Python function that reverses a string.
73 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-seed-run1 stop-seed-run1 OK 53 0.7 1.0 1135 9.66
117.5
Name three primary colors and explain why they matter in des...
74 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-system-numbered stop-system-numbered OK 23 0.0 1.0 1455 12.4
117.4
Name three primary colors and explain why they matter in des...
75 mlx-community/Qwen3.5-35B-A3B-4bit @ top-p top-p
--top-p 0.9
OK 33 0.8 1.0 912 7.78
117.3
Explain why the sky is blue in exactly 3 bullet points.
76 mlx-community/Qwen3.5-35B-A3B-4bit @ json-output json-output OK 43 0.0 1.0 810 6.91
117.3
Name three primary colors and explain why they matter in des...
77 mlx-community/Qwen3.5-35B-A3B-4bit @ long-output long-output OK 23 0.7 1.0 1994 17.04
117.0
Write a detailed recipe for chocolate chip cookies with ingr...
78 mlx-community/Qwen3.5-35B-A3B-4bit @ strict-format strict-format OK 33 0.0 1.0 334 2.86
117.0
Output EXACTLY 3 lines. Each line must contain exactly one w...
79 mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-multi tool-call-multi OK 33 0.0 1.0 1169 10.0
116.9
What is the weather in London and what time is it in Tokyo?
80 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-no-match stop-no-match OK 23 0.0 1.0 256 2.19
116.9
Explain the difference between TCP and UDP in a few sentence...
81 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-seed-run2 stop-seed-run2 OK 53 0.7 1.0 1135 9.71
116.9
Name three primary colors and explain why they matter in des...
82 mlx-community/Qwen3.5-35B-A3B-4bit @ code-swift code-swift OK 33 0.0 1.0 4096 35.07
116.8
Write a Swift async function that fetches JSON from a URL us...
83 mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-complex tool-call-complex OK 33 0.0 1.0 338 2.9
116.7
tools: [{"type":"function","function":{"name":"search_files"...
84 mlx-community/Qwen3.5-35B-A3B-4bit @ long-form long-form OK 33 0.7 1.0 1536 13.16
116.7
Name three primary colors and explain why they matter in des...
85 mlx-community/Qwen3.5-35B-A3B-4bit @ numbered-list numbered-list OK 53 0.0 1.0 810 6.94
116.7
Name three primary colors and explain why they matter in des...
86 mlx-community/Qwen3.5-35B-A3B-4bit @ min-p min-p
--min-p 0.05
OK 53 0.8 1.0 1249 10.72
116.5
Name three primary colors and explain why they matter in des...
87 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-system-numbered stop-system-numbered OK 23 0.0 1.0 874 7.5
116.5
What are the main benefits of exercise?
88 mlx-community/Qwen3.5-35B-A3B-4bit @ response-format-json response-format-json OK 23 0.0 1.0 810 6.98
116.0
Name three primary colors and explain why they matter in des...
89 mlx-community/Qwen3.5-35B-A3B-4bit @ think-raw think-raw
--raw
OK 33 0.0 1.0 810 7.01
115.5
Name three primary colors and explain why they matter in des...
90 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-json-object-key stop-json-object-key OK 23 0.0 1.0 851 7.37
115.5
Name three primary colors and explain why they matter in des...
91 mlx-community/Qwen3.5-35B-A3B-4bit @ guided-json-simple guided-json-simple
--guided-json '{"type":"object","properties":{"name":{"type":"string"},"age":{"type":"integer"}},"required":["name","age"]}'
OK 53 0.0 1.0 810 7.02
115.4
Name three primary colors and explain why they matter in des...
92 mlx-community/Qwen3.5-35B-A3B-4bit @ prefill-default prefill-default OK 53 0.0 1.0 2610 22.61
115.4
Name three primary colors and explain why they matter in des...
93 mlx-community/Qwen3.5-35B-A3B-4bit @ response-format-text response-format-text OK 33 0.0 1.0 810 7.02
115.3
Name three primary colors and explain why they matter in des...
94 mlx-community/Qwen3.5-35B-A3B-4bit @ agent-cached-turn2 agent-cached-turn2
--enable-prefix-caching
OK 43 0.0 1.0 134 1.16
115.3
Now add a --timeout flag to the CLI that sets a request time...
95 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-multi stop-multi OK 43 0.0 1.0 810 7.02
115.3
Name three primary colors and explain why they matter in des...
96 mlx-community/Qwen3.5-35B-A3B-4bit @ combined-samplers combined-samplers
--top-k 50 --min-p 0.03 --top-p 0.95
OK 43 0.8 1.0 632 5.48
115.3
Explain why the sky is blue in exactly 3 bullet points.
97 mlx-community/Qwen3.5-35B-A3B-4bit @ response-format-schema response-format-schema OK 53 0.0 1.0 810 7.03
115.2
Name three primary colors and explain why they matter in des...
98 mlx-community/Qwen3.5-35B-A3B-4bit @ developer-role developer-role OK 33 0.0 1.0 810 7.04
115.1
Name three primary colors and explain why they matter in des...
99 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-cli-api-dedup stop-cli-api-dedup
--stop "3."
OK 23 0.0 1.0 810 7.05
115.0
Name three primary colors and explain why they matter in des...
100 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-cli-multi stop-cli-multi
--stop "```,DONE"
OK 23 0.0 1.0 810 7.05
114.8
Name three primary colors and explain why they matter in des...
101 mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-auto tool-call-auto OK 33 0.0 1.0 330 2.87
114.8
tools: [{"type":"function","function":{"name":"get_weather",...
102 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-special-chars stop-special-chars OK 23 0.0 2.0 824 7.18
114.8
Name three primary colors and explain why they matter in des...
103 mlx-community/Qwen3.5-35B-A3B-4bit @ verbose verbose
--verbose
OK 33 0.7 1.0 824 7.18
114.8
Name three primary colors and explain why they matter in des...
104 mlx-community/Qwen3.5-35B-A3B-4bit @ code-swift code-swift OK 33 0.0 1.0 4096 35.73
114.6
Name three primary colors and explain why they matter in des...
105 mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-none tool-call-none OK 33 0.0 1.0 810 7.07
114.5
Name three primary colors and explain why they matter in des...
106 mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-auto tool-call-auto OK 33 0.0 1.0 810 7.07
114.5
Name three primary colors and explain why they matter in des...
107 mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-xml tool-call-xml
--tool-call-parser xmlFunction
OK 33 0.0 1.0 330 2.88
114.5
tools: [{"type":"function","function":{"name":"get_weather",...
108 mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-complex tool-call-complex OK 33 0.0 1.0 810 7.08
114.3
Name three primary colors and explain why they matter in des...
109 mlx-community/Qwen3.5-35B-A3B-4bit @ long-output long-output OK 23 0.7 1.0 1172 10.26
114.2
Name three primary colors and explain why they matter in des...
110 mlx-community/Qwen3.5-35B-A3B-4bit @ max-completion-tokens max-completion-tokens OK 33 0.0 1.0 810 7.1
114.1
Name three primary colors and explain why they matter in des...
111 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-single stop-single OK 23 0.0 1.0 810 7.1
114.1
Name three primary colors and explain why they matter in des...
112 mlx-community/Qwen3.5-35B-A3B-4bit @ long-form long-form OK 33 0.7 1.0 3837 33.71
113.8
Write a detailed technical blog post explaining how Mixture-...
113 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-unicode stop-unicode OK 53 0.0 1.0 810 7.13
113.6
Name three primary colors and explain why they matter in des...
114 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-cli-only stop-cli-only
--stop "3."
OK 53 0.0 1.0 810 7.14
113.4
Name three primary colors and explain why they matter in des...
115 mlx-community/Qwen3.5-35B-A3B-4bit @ greedy greedy OK 23 0.0 1.0 810 7.15
113.3
Name three primary colors and explain why they matter in des...
116 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-multi-word stop-multi-word OK 23 0.0 1.0 810 7.15
113.3
Name three primary colors and explain why they matter in des...
117 mlx-community/Qwen3.5-35B-A3B-4bit @ special-chars special-chars OK 33 0.0 2.0 4028 35.63
113.1
Repeat these characters exactly: <tag> "quotes" 'apostrophes...
118 mlx-community/Qwen3.5-35B-A3B-4bit @ think-normal think-normal OK 33 0.0 1.0 810 7.16
113.0
Name three primary colors and explain why they matter in des...
119 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-word stop-word OK 33 0.0 1.0 1214 10.75
113.0
Name 5 programming languages and briefly describe each one.
120 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-guided-json-brace stop-guided-json-brace
--guided-json '{"type":"object","properties":{"color":{"type":"string"},"hex":{"type":"string"}},"required":["color","hex"]}'
OK 23 0.0 1.0 810 7.17
112.9
Name three primary colors and explain why they matter in des...
121 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-cli-api-merge stop-cli-api-merge
--stop "5."
OK 23 0.0 1.0 810 7.18
112.8
Name three primary colors and explain why they matter in des...
122 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-guided-json-value stop-guided-json-value
--guided-json '{"type":"object","properties":{"cities":{"type":"array","items":{"type":"string"}}},"required":["cities"]}'
OK 23 0.0 1.0 810 7.18
112.8
Name three primary colors and explain why they matter in des...
123 mlx-community/Qwen3.5-35B-A3B-4bit @ raw-mode raw-mode
--raw
OK 23 0.7 1.0 724 6.42
112.8
Name three primary colors and explain why they matter in des...
124 mlx-community/Qwen3.5-35B-A3B-4bit @ small-kv small-kv
--max-kv-size 2048
OK 23 0.7 1.0 1223 10.86
112.7
Name three primary colors and explain why they matter in des...
125 mlx-community/Qwen3.5-35B-A3B-4bit @ default default OK 53 0.7 1.0 1003 8.91
112.6
Name three primary colors and explain why they matter in des...
126 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-guided-json-comma stop-guided-json-comma
--guided-json '{"type":"object","properties":{"name":{"type":"string"},"age":{"type":"integer"},"city":{"type":"string"}},"required":["name","age","city"]}'
OK 23 0.0 1.0 763 6.78
112.5
Generate a person profile for someone named Alice who is 30 ...
127 mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-xml tool-call-xml
--tool-call-parser xmlFunction
OK 33 0.0 1.0 810 7.2
112.4
Name three primary colors and explain why they matter in des...
128 mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-multi tool-call-multi OK 33 0.0 1.0 287 2.55
112.4
tools: [{"type":"function","function":{"name":"get_weather",...
129 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-html-tag stop-html-tag OK 53 0.0 2.0 810 7.24
111.8
Name three primary colors and explain why they matter in des...
130 mlx-community/Qwen3.5-35B-A3B-4bit @ logprobs logprobs
--max-logprobs 5
OK 23 0.0 1.0 810 7.24
111.8
Name three primary colors and explain why they matter in des...
131 mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-multi tool-call-multi OK 33 0.0 1.0 810 7.26
111.6
Name three primary colors and explain why they matter in des...
132 mlx-community/Qwen3.5-35B-A3B-4bit @ strict-format strict-format OK 33 0.0 1.0 810 7.26
111.6
Name three primary colors and explain why they matter in des...
133 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-cli-multi stop-cli-multi
--stop "```,DONE"
OK 43 0.0 1.0 83 0.74
111.6
Write a bash script that prints the current date inside a co...
134 mlx-community/Qwen3.5-35B-A3B-4bit @ minimal-prompt minimal-prompt OK 33 0.7 1.0 645 5.79
111.4
Name three primary colors and explain why they matter in des...
135 mlx-community/Qwen3.5-35B-A3B-4bit @ long-prompt long-prompt OK 33 0.0 1.0 810 7.28
111.2
Name three primary colors and explain why they matter in des...
136 mlx-community/Qwen3.5-35B-A3B-4bit @ kv-quantized kv-quantized
--kv-bits 4
OK 53 0.7 2.0 1003 9.02
111.2
Name three primary colors and explain why they matter in des...
137 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-word stop-word OK 23 0.0 1.0 810 7.29
111.2
Name three primary colors and explain why they matter in des...
138 mlx-community/Qwen3.5-35B-A3B-4bit @ math math OK 33 0.0 2.0 810 7.31
110.9
Name three primary colors and explain why they matter in des...
139 mlx-community/Qwen3.5-35B-A3B-4bit @ top-p top-p
--top-p 0.9
OK 53 0.8 1.0 976 8.83
110.5
Name three primary colors and explain why they matter in des...
140 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-non-streaming stop-non-streaming
--no-streaming
OK 53 0.0 1.0 810 7.33
110.5
Name three primary colors and explain why they matter in des...
141 mlx-community/Qwen3.5-35B-A3B-4bit @ code-python code-python OK 33 0.0 2.0 267 2.42
110.3
Write a function that finds all prime numbers up to n using ...
142 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-long-phrase stop-long-phrase OK 23 0.0 1.0 810 7.35
110.2
Name three primary colors and explain why they matter in des...
143 mlx-community/Qwen3.5-35B-A3B-4bit @ combined-samplers combined-samplers
--top-k 50 --min-p 0.03 --top-p 0.95
OK 43 0.8 1.0 772 7.01
110.1
Name three primary colors and explain why they matter in des...
144 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-low-max-tokens stop-low-max-tokens OK 23 0.0 2.0 100 0.92
108.8
List 10 mountains, numbered 1 through 10, one per line.
145 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-multi stop-multi OK 43 0.0 1.0 253 2.34
108.3
Write a Python hello world program in a code block, then wri...
146 mlx-community/Qwen3.5-35B-A3B-4bit @ logprobs logprobs
--max-logprobs 5
OK 23 0.0 1.0 144 1.33
108.2
What is 1+1?
147 mlx-community/Qwen3.5-35B-A3B-4bit @ agent-no-cache-turn3 agent-no-cache-turn3 OK 23 0.0 1.0 149 1.38
107.7
Write a unit test for the timeout feature you just described...
148 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-unicode stop-unicode OK 43 0.0 1.0 341 3.17
107.5
List 5 items about space using bullet points (β€’).
149 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-special-chars stop-special-chars OK 53 0.0 2.0 558 5.3
105.3
List 3 facts about the moon. Use **bold** markdown for empha...
150 mlx-community/Qwen3.5-35B-A3B-4bit @ special-chars special-chars OK 33 0.0 2.0 810 7.71
105.1
Name three primary colors and explain why they matter in des...
151 mlx-community/Qwen3.5-35B-A3B-4bit @ top-k top-k
--top-k 30
OK 43 0.8 2.0 810 7.72
104.9
Name three primary colors and explain why they matter in des...
152 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-system-pirate stop-system-pirate OK 23 0.0 1.0 685 6.61
103.7
Tell me about treasure hunting on the high seas.
153 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-newline stop-newline OK 23 0.0 1.0 173 1.67
103.5
What is the capital of France? Answer in one sentence.
154 mlx-community/Qwen3.5-35B-A3B-4bit @ multilingual multilingual OK 33 0.3 1.0 640 6.24
102.6
Name three primary colors and explain why they matter in des...
155 mlx-community/Qwen3.5-35B-A3B-4bit @ agent-cached-turn1 agent-cached-turn1
--enable-prefix-caching
OK 43 0.0 1.0 59 0.58
101.5
Read the file Sources/MacLocalAPI/main.swift and explain wha...
156 mlx-community/Qwen3.5-35B-A3B-4bit @ short-output short-output OK 23 0.7 1.0 50 0.5
100.0
Describe the entire history of Rome.
157 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-no-match stop-no-match OK 53 0.0 1.0 256 2.56
100.0
Name three primary colors and explain why they matter in des...
158 mlx-community/Qwen3.5-35B-A3B-4bit @ code-python code-python OK 33 0.0 2.0 4096 41.18
99.5
Name three primary colors and explain why they matter in des...
159 mlx-community/Qwen3.5-35B-A3B-4bit @ agent-cached-turn3 agent-cached-turn3
--enable-prefix-caching
OK 23 0.0 1.0 256 2.61
98.2
Name three primary colors and explain why they matter in des...
160 mlx-community/Qwen3.5-35B-A3B-4bit @ agent-no-cache-turn1 agent-no-cache-turn1 OK 23 0.0 1.0 256 2.61
98.1
Name three primary colors and explain why they matter in des...
161 mlx-community/Qwen3.5-35B-A3B-4bit @ repetition-penalty repetition-penalty
--repetition-penalty 1.2
OK 33 0.8 1.0 3488 35.61
98.0
Write a long essay about the history of bread making across ...
162 mlx-community/Qwen3.5-35B-A3B-4bit @ agent-no-cache-turn2 agent-no-cache-turn2 OK 23 0.0 1.0 256 2.61
98.0
Name three primary colors and explain why they matter in des...
163 mlx-community/Qwen3.5-35B-A3B-4bit @ with-penalty with-penalty
--presence-penalty 1.5
OK 43 0.8 2.0 4096 41.83
97.9
Write a long essay about the history of bread making across ...
164 mlx-community/Qwen3.5-35B-A3B-4bit @ agent-no-cache-turn3 agent-no-cache-turn3 OK 33 0.0 1.0 256 2.62
97.8
Name three primary colors and explain why they matter in des...
165 mlx-community/Qwen3.5-35B-A3B-4bit @ agent-no-cache-turn2 agent-no-cache-turn2 OK 43 0.0 1.0 103 1.05
97.8
Now add a --timeout flag to the CLI that sets a request time...
166 mlx-community/Qwen3.5-35B-A3B-4bit @ agent-cached-turn1 agent-cached-turn1
--enable-prefix-caching
OK 23 0.0 1.0 256 2.62
97.7
Name three primary colors and explain why they matter in des...
167 mlx-community/Qwen3.5-35B-A3B-4bit @ multilingual multilingual OK 33 0.3 1.0 714 7.33
97.4
Translate "Hello, how are you?" into French, Spanish, Japane...
168 mlx-community/Qwen3.5-35B-A3B-4bit @ repetition-penalty repetition-penalty
--repetition-penalty 1.2
OK 53 0.8 1.0 1853 19.32
95.9
Name three primary colors and explain why they matter in des...
169 mlx-community/Qwen3.5-35B-A3B-4bit @ agent-cached-turn2 agent-cached-turn2
--enable-prefix-caching
OK 23 0.0 1.0 256 2.68
95.7
Name three primary colors and explain why they matter in des...
170 mlx-community/Qwen3.5-35B-A3B-4bit @ with-penalty with-penalty
--presence-penalty 1.5
OK 53 0.8 2.0 1582 16.61
95.2
Name three primary colors and explain why they matter in des...
171 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-multi-word stop-multi-word OK 33 0.0 1.0 374 3.97
94.2
Write a 5-step recipe for making tea. Label each step as "St...
172 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-guided-json-value stop-guided-json-value
--guided-json '{"type":"object","properties":{"cities":{"type":"array","items":{"type":"string"}}},"required":["cities"]}'
OK 23 0.0 1.0 209 2.23
93.9
List 5 major world cities as a JSON array. Include Tokyo.
173 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-cli-api-dedup stop-cli-api-dedup
--stop "3."
OK 23 0.0 1.0 234 2.57
91.1
List 10 cities, numbered 1 through 10, one per line.
174 mlx-community/Qwen3.5-35B-A3B-4bit @ agent-no-cache-turn1 agent-no-cache-turn1 OK 23 0.0 1.0 67 0.75
89.9
Read the file Sources/MacLocalAPI/main.swift and explain wha...
175 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-seed-run2 stop-seed-run2 OK 23 0.7 1.0 329 3.67
89.7
List 10 flowers, numbered 1 through 10, one per line.
176 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-seed-run1 stop-seed-run1 OK 23 0.7 1.0 329 3.68
89.3
List 10 flowers, numbered 1 through 10, one per line.
177 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-cli-api-merge stop-cli-api-merge
--stop "5."
OK 23 0.0 1.0 245 2.82
86.7
List 10 countries, numbered 1 through 10, one per line.
178 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-single stop-single OK 23 0.0 1.0 199 2.38
83.6
List 10 fruits, numbered 1-10, one per line.
179 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-high-temp stop-high-temp OK 23 1.0 1.0 605 7.35
82.3
List 10 random words, numbered 1 through 10, one per line.
180 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-cli-only stop-cli-only
--stop "3."
OK 23 0.0 1.0 209 2.56
81.6
List 10 types of cheese, numbered 1 through 10, one per line...
181 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-html-tag stop-html-tag OK 23 0.0 2.0 232 2.87
80.9
Write an HTML unordered list of 5 fruits using <ul> and <li>...
182 mlx-community/Qwen3.5-35B-A3B-4bit @ long-prompt long-prompt OK 33 0.0 1.0 831 10.36
80.2
Repeat after me exactly: The quick brown fox jumps over the ...
183 mlx-community/Qwen3.5-35B-A3B-4bit @ stop-low-max-tokens stop-low-max-tokens OK 43 0.0 2.0 100 1.44
69.4
Name three primary colors and explain why they matter in des...
184 mlx-community/Qwen3.5-35B-A3B-4bit @ short-output short-output OK 23 0.7 1.0 50 0.89
56.3
Name three primary colors and explain why they matter in des...

AI Analysis (--smart)

codex Analysis · avg score: 3.0/5

claude Analysis · avg score: 3.0/5

Full Responses

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-long-phrase 1143 tokens · 145.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-double-newline 742 tokens · 128.0 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-immediate 767 tokens · 125.6 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-newline 817 tokens · 124.8 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-period 692 tokens · 124.7 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-period 831 tokens · 124.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ guided-json-simple 1373 tokens · 124.1 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ seed-42-run2 4096 tokens · 123.6 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ no-penalty 3795 tokens · 123.4 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ scientist 2402 tokens · 123.4 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ guided-json-nested 937 tokens · 123.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-double-newline 817 tokens · 123.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ response-format-json 1303 tokens · 123.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-four-max 4096 tokens · 123.2 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ kv-quantized 1132 tokens · 123.1 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ developer-role 4096 tokens · 123.1 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-immediate 817 tokens · 123.0 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-four-max 818 tokens · 123.0 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ eli5 921 tokens · 122.8 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ small-kv 1184 tokens · 122.7 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ json-output 384 tokens · 122.6 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ think-raw 833 tokens · 122.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ response-format-schema 1478 tokens · 122.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ seed-42-run1 4096 tokens · 122.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ non-streaming-seeded 2469 tokens · 122.4 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ eli5 4096 tokens · 122.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ agent-cached-turn3 356 tokens · 122.1 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ prefill-small-256 3652 tokens · 122.1 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ max-completion-tokens 2300 tokens · 121.9 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ max-completion-tokens 1174 tokens · 121.7 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ no-streaming 4096 tokens · 121.6 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ greedy 546 tokens · 121.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ prefill-large-4096 3652 tokens · 121.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ min-p 793 tokens · 121.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-guided-json-comma 826 tokens · 121.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ streaming-seeded 2469 tokens · 121.2 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ raw-mode 4096 tokens · 121.2 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-xml 4096 tokens · 121.1 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ math 1830 tokens · 121.0 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ pirate 1459 tokens · 121.0 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ think-normal 833 tokens · 121.0 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ pirate 1836 tokens · 121.0 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-json-object-key 4096 tokens · 120.9 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ very-verbose 658 tokens · 120.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ numbered-list 489 tokens · 120.4 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ prefill-small-256 2610 tokens · 120.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-none 475 tokens · 120.2 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-system-pirate 4096 tokens · 120.2 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ no-penalty 1580 tokens · 120.0 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ high-temp 1357 tokens · 119.9 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ prefill-default 3652 tokens · 119.9 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-non-streaming 4096 tokens · 119.8 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-guided-json-brace 1063 tokens · 119.8 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-high-temp 1771 tokens · 119.6 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ response-format-text 370 tokens · 119.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-complex 458 tokens · 119.4 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ scientist 2950 tokens · 119.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ high-temp 852 tokens · 119.1 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ no-streaming 1623 tokens · 119.1 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ seed-42-run1 1135 tokens · 118.8 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ seed-42-run2 1135 tokens · 118.8 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ streaming-seeded 1501 tokens · 118.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-auto 502 tokens · 118.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ verbose 387 tokens · 118.4 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ very-verbose 1894 tokens · 118.4 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ default 1090 tokens · 118.4 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ prefill-large-4096 2610 tokens · 118.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ minimal-prompt 390 tokens · 118.2 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ non-streaming-seeded 1501 tokens · 118.0 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ top-k 625 tokens · 117.9 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ guided-json-nested 810 tokens · 117.7 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ developer-role 437 tokens · 117.6 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-seed-run1 1135 tokens · 117.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-system-numbered 1455 tokens · 117.4 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ top-p 912 tokens · 117.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ json-output 810 tokens · 117.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ long-output 1994 tokens · 117.0 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ strict-format 334 tokens · 117.0 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-multi 1169 tokens · 116.9 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-no-match 256 tokens · 116.9 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-seed-run2 1135 tokens · 116.9 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ code-swift 4096 tokens · 116.8 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-complex 338 tokens · 116.7 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ long-form 1536 tokens · 116.7 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ numbered-list 810 tokens · 116.7 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ min-p 1249 tokens · 116.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-system-numbered 874 tokens · 116.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ response-format-json 810 tokens · 116.0 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ think-raw 810 tokens · 115.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-json-object-key 851 tokens · 115.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ guided-json-simple 810 tokens · 115.4 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ prefill-default 2610 tokens · 115.4 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ response-format-text 810 tokens · 115.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ agent-cached-turn2 134 tokens · 115.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-multi 810 tokens · 115.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ combined-samplers 632 tokens · 115.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ response-format-schema 810 tokens · 115.2 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ developer-role 810 tokens · 115.1 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-cli-api-dedup 810 tokens · 115.0 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-cli-multi 810 tokens · 114.8 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-auto 330 tokens · 114.8 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-special-chars 824 tokens · 114.8 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ verbose 824 tokens · 114.8 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ code-swift 4096 tokens · 114.6 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-none 810 tokens · 114.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-auto 810 tokens · 114.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-xml 330 tokens · 114.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-complex 810 tokens · 114.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ long-output 1172 tokens · 114.2 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ max-completion-tokens 810 tokens · 114.1 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-single 810 tokens · 114.1 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ long-form 3837 tokens · 113.8 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-unicode 810 tokens · 113.6 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-cli-only 810 tokens · 113.4 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ greedy 810 tokens · 113.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-multi-word 810 tokens · 113.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ special-chars 4028 tokens · 113.1 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ think-normal 810 tokens · 113.0 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-word 1214 tokens · 113.0 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-guided-json-brace 810 tokens · 112.9 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-cli-api-merge 810 tokens · 112.8 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-guided-json-value 810 tokens · 112.8 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ raw-mode 724 tokens · 112.8 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ small-kv 1223 tokens · 112.7 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ default 1003 tokens · 112.6 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-guided-json-comma 763 tokens · 112.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-xml 810 tokens · 112.4 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-multi 287 tokens · 112.4 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-html-tag 810 tokens · 111.8 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ logprobs 810 tokens · 111.8 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ tool-call-multi 810 tokens · 111.6 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ strict-format 810 tokens · 111.6 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-cli-multi 83 tokens · 111.6 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ minimal-prompt 645 tokens · 111.4 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ long-prompt 810 tokens · 111.2 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ kv-quantized 1003 tokens · 111.2 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-word 810 tokens · 111.2 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ math 810 tokens · 110.9 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ top-p 976 tokens · 110.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-non-streaming 810 tokens · 110.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ code-python 267 tokens · 110.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-long-phrase 810 tokens · 110.2 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ combined-samplers 772 tokens · 110.1 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-low-max-tokens 100 tokens · 108.8 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-multi 253 tokens · 108.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ logprobs 144 tokens · 108.2 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ agent-no-cache-turn3 149 tokens · 107.7 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-unicode 341 tokens · 107.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-special-chars 558 tokens · 105.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ special-chars 810 tokens · 105.1 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ top-k 810 tokens · 104.9 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-system-pirate 685 tokens · 103.7 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-newline 173 tokens · 103.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ multilingual 640 tokens · 102.6 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ agent-cached-turn1 59 tokens · 101.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ short-output 50 tokens · 100.0 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-no-match 256 tokens · 100.0 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ code-python 4096 tokens · 99.5 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ agent-cached-turn3 256 tokens · 98.2 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ agent-no-cache-turn1 256 tokens · 98.1 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ repetition-penalty 3488 tokens · 98.0 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ agent-no-cache-turn2 256 tokens · 98.0 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ with-penalty 4096 tokens · 97.9 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ agent-no-cache-turn3 256 tokens · 97.8 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ agent-no-cache-turn2 103 tokens · 97.8 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ agent-cached-turn1 256 tokens · 97.7 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ multilingual 714 tokens · 97.4 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ repetition-penalty 1853 tokens · 95.9 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ agent-cached-turn2 256 tokens · 95.7 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ with-penalty 1582 tokens · 95.2 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-multi-word 374 tokens · 94.2 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-guided-json-value 209 tokens · 93.9 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-cli-api-dedup 234 tokens · 91.1 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ agent-no-cache-turn1 67 tokens · 89.9 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-seed-run2 329 tokens · 89.7 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-seed-run1 329 tokens · 89.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-cli-api-merge 245 tokens · 86.7 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-single 199 tokens · 83.6 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-high-temp 605 tokens · 82.3 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-cli-only 209 tokens · 81.6 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-html-tag 232 tokens · 80.9 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ long-prompt 831 tokens · 80.2 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ stop-low-max-tokens 100 tokens · 69.4 tok/s

mlx-community/Qwen3.5-35B-A3B-4bit @ short-output 50 tokens · 56.3 tok/s