mistral-small-3.1-24b-instruct-2503

evaluatedReasoner

Vendor: Mistral AIParams: 24BQuant: Q8Context: 128kLicense: Apache-2.0Type: local

Capability starfish

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

Overall

mean domain rate

Solved

43/59

latest per bench

Gauntlet time

~8s/bench

Speed

8.2

tok/s

Steps/bench

1.7

decisiveness

$ / run

$0.0004

OpenRouter-equiv

Overall rank

#14 of 26

Agentic

#11 of 22

Coding

#14 of 22

Speed (tok/s)

#18 of 22

Decisive

#15 of 22

Cheapest

#9 of 9

Strongest

Gate · 100%

Weakest

C# · 80%

Where it fits · live, recomputed from current results

#14 of 26 overall; strongest on Gate (100%); weakest on C# (80%); #18/22 for speed (8.2 tok/s); #9/9 cheapest ($0.0004/run). (auto-generated from current results)

curator note: A deliberate reasoner — burns the most wall-clock per bench (~5-6h for the gauntlet) but thinks hard. Not for latency-sensitive loops; useful where a slow, careful pass pays off.

densereasonerslow

Verdict history · snapshotted daily when it changes

2026-07-09#14 of 26 overall; strongest on Gate (100%); weakest on C# (80%); #18/22 for speed (8.2 tok/s); #9/9 cheapest ($0.0004/run). (auto-generated from current results)

2026-07-08#15 of 26 overall; strongest on Gate (100%); weakest on C# (80%); #18/22 for speed (8.2 tok/s); #9/9 cheapest ($0.0004/run). (auto-generated from current results)

2026-07-07#18 of 28 overall; strongest on Gate (100%); weakest on SupportFlow (14%); #18/25 on the supportflow gauntlet; #19/26 for speed (7.9 tok/s); #9/9 cheapest ($0.0241/run). (auto-generated from current results)

2026-07-06#17 of 28 overall; strongest on Gate (100%); weakest on SupportFlow (14%); #18/24 on the supportflow gauntlet; #19/26 for speed (7.9 tok/s); #9/9 cheapest ($0.0241/run). (auto-generated from current results)

2026-07-05#17 of 28 overall; strongest on Gate (100%); weakest on SupportFlow (14%); #19/23 on the supportflow gauntlet; #19/26 for speed (7.9 tok/s); #9/9 cheapest ($0.0241/run). (auto-generated from current results)

2026-07-04#17 of 28 overall; strongest on Gate (100%); weakest on SupportFlow (16%); #18/23 on the supportflow gauntlet; #19/26 for speed (7.9 tok/s); #9/9 cheapest ($0.0221/run). (auto-generated from current results)

2026-07-03#17 of 28 overall; strongest on Gate (100%); weakest on SupportFlow (15%); #17/22 on the supportflow gauntlet; #19/26 for speed (8.0 tok/s); #9/9 cheapest ($0.0198/run). (auto-generated from current results)

2026-07-02#20 of 30 overall; strongest on Gate (100%); weakest on PHP (0%); #19/26 on the supportflow gauntlet; #21/29 for speed (8.0 tok/s); #9/9 cheapest ($0.0198/run). (auto-generated from current results)

2026-07-01#12 of 28 overall; strongest on Gate (100%); weakest on SupportFlow (15%); #16/23 on the supportflow gauntlet; #20/27 for speed (8.0 tok/s); #9/9 cheapest ($0.0198/run). (auto-generated from current results)

2026-06-30#12 of 28 overall; strongest on Gate (100%); weakest on SupportFlow (15%); #15/21 on the supportflow gauntlet; #20/27 for speed (8.0 tok/s); #9/9 cheapest ($0.0198/run). (auto-generated from current results)

2026-06-29#11 of 28 overall; strongest on Gate (100%); weakest on SupportFlow (15%); #15/21 on the supportflow gauntlet; #20/27 for speed (8.0 tok/s); #9/9 cheapest ($0.0198/run). (auto-generated from current results)

Domain breakdown

Gate

100%

14/14

Coding

83%

5/6

Agentic

89%

8/9

80%

12/15

PHP

80%

4/5

Benches run

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

Gate (14)

Coding (6)

Agentic (9)

C# (15)

PHP (5)

SWE (10)

History (monthly pass-rate)

2026-06

92%

209/228

Availability record

When this model appeared in our catalogue census.

2026-06-25 · local