Touchstone
← Models

deepseek-r1-distill-qwen-32b

evaluated
Vendor: DeepSeek
Capability starfish
green = capability (pass-rate) · blue = operational (vs roster) · dashed = roster average
Gate7/11Coding6/6Agentic4/11C#10/15PHP0/5Speed5.0 tok/sDecisive1.2 steps/benchCost-effn/a0
53%
Overall
mean domain rate
Solved
27/48
latest per bench
Gauntlet time
49m
~105s/bench
Speed
5
tok/s
Steps/bench
1.2
decisiveness
$ / run
OpenRouter-equiv
Overall rank
#22 of 26
Agentic
#22 of 22
Coding
#8 of 22
Speed (tok/s)
#21 of 22
Decisive
#1 of 22
Strongest
Coding · 100%
Weakest
PHP · 0%
Where it fits · live, recomputed from current results

#22 of 26 overall; strongest on Coding (100%); weakest on PHP (0%); #21/22 for speed (5.0 tok/s). (auto-generated from current results)

Verdict history · snapshotted daily when it changes
2026-07-08#22 of 26 overall; strongest on Coding (100%); weakest on PHP (0%); #21/22 for speed (5.0 tok/s). (auto-generated from current results)
2026-07-07#23 of 28 overall; strongest on Coding (100%); weakest on PHP (0%); #16/25 on the supportflow gauntlet; #23/26 for speed (5.4 tok/s). (auto-generated from current results)
2026-07-06#22 of 28 overall; strongest on Coding (100%); weakest on PHP (0%); #16/24 on the supportflow gauntlet; #23/26 for speed (5.4 tok/s). (auto-generated from current results)
2026-07-05#22 of 28 overall; strongest on Coding (100%); weakest on PHP (0%); #9/23 on the supportflow gauntlet; #24/26 for speed (5.2 tok/s). (auto-generated from current results)
2026-07-04#22 of 28 overall; strongest on Coding (100%); weakest on PHP (0%); #11/23 on the supportflow gauntlet; #24/26 for speed (5.2 tok/s). (auto-generated from current results)
2026-07-03#24 of 28 overall; strongest on Gate (64%); weakest on PHP (0%); #24/26 for speed (4.7 tok/s). (auto-generated from current results)
2026-07-02#24 of 30 overall; strongest on Gate (64%); weakest on SupportFlow (0%); #25/26 on the supportflow gauntlet; #27/29 for speed (4.7 tok/s). (auto-generated from current results)
2026-07-01#25 of 28 overall; strongest on Gate (27%); weakest on Coding (0%); #22/23 on the supportflow gauntlet; #23/27 for speed (6.6 tok/s). (auto-generated from current results)
2026-06-29#27 of 28 overall; strongest on SupportFlow (0%); weakest on SupportFlow (0%); #20/21 on the supportflow gauntlet; #24/27 for speed (6.2 tok/s). (auto-generated from current results)
Domain breakdown
Gate
64%
7/11
Coding
100%
6/6
Agentic
36%
4/11
C#
67%
10/15
PHP
0%
0/5
Benches run
green = passed · red = failed · hover for harness & time
Gate (11)
Coding (6)
Agentic (11)
C# (15)
PHP (5)
History (monthly pass-rate)
2026-07
59%
79/134
Availability record
When this model appeared in our catalogue census.

No census snapshots recorded yet.