qwen3.5-2b

evaluated

Vendor: AlibabaQuant: Q8Type: local

Capability starfish

green = capability (pass-rate) · blue = operational (vs roster) · dashed = roster average

Overall

mean domain rate

Solved

31/54

latest per bench

Gauntlet time

~1s/bench

Speed

68.9

tok/s

Steps/bench

decisiveness

$ / run

—

OpenRouter-equiv

Overall rank

#23 of 26

Agentic

#15 of 22

Coding

#21 of 22

Speed (tok/s)

#4 of 22

Decisive

#19 of 22

Strongest

Agentic · 86%

Weakest

Coding · 33%

Where it fits · live, recomputed from current results

#23 of 26 overall; strongest on Agentic (86%); weakest on Coding (33%); #4/22 for speed (68.9 tok/s). (auto-generated from current results)

Verdict history · snapshotted daily when it changes

2026-07-08#23 of 26 overall; strongest on Agentic (86%); weakest on Coding (33%); #4/22 for speed (68.9 tok/s). (auto-generated from current results)

2026-07-07#24 of 28 overall; strongest on Agentic (86%); weakest on SupportFlow (0%); #23/25 on the supportflow gauntlet; #4/26 for speed (60.0 tok/s). (auto-generated from current results)

2026-07-06#24 of 28 overall; strongest on Agentic (86%); weakest on SupportFlow (0%); #22/24 on the supportflow gauntlet; #4/26 for speed (60.0 tok/s). (auto-generated from current results)

2026-07-04#24 of 28 overall; strongest on Agentic (86%); weakest on SupportFlow (0%); #22/23 on the supportflow gauntlet; #4/26 for speed (60.1 tok/s). (auto-generated from current results)

2026-07-03#23 of 28 overall; strongest on Agentic (86%); weakest on SupportFlow (0%); #21/22 on the supportflow gauntlet; #4/26 for speed (60.1 tok/s). (auto-generated from current results)

2026-07-02#23 of 30 overall; strongest on Agentic (86%); weakest on SupportFlow (3%); #22/26 on the supportflow gauntlet; #9/29 for speed (35.7 tok/s). (auto-generated from current results)

2026-07-01#19 of 28 overall; strongest on Agentic (86%); weakest on SupportFlow (0%); #19/23 on the supportflow gauntlet; #3/27 for speed (59.4 tok/s). (auto-generated from current results)

2026-06-30#18 of 28 overall; strongest on Agentic (78%); weakest on PHP (20%); #3/27 for speed (67.0 tok/s). (auto-generated from current results)

2026-06-29#17 of 28 overall; strongest on Agentic (78%); weakest on PHP (20%); #3/27 for speed (67.0 tok/s). (auto-generated from current results)

Domain breakdown

Gate

64%

9/14

Coding

33%

2/6

Agentic

86%

12/14

40%

6/15

PHP

40%

2/5

Benches run

green = passed · red = failed · hover for harness & time

Gate (14)

Coding (6)

Agentic (14)

C# (15)

PHP (5)

History (monthly pass-rate)

2026-06

66%

225/340

2026-07

81%

188/232

Availability record

When this model appeared in our catalogue census.

No census snapshots recorded yet.