Structure & Data · blind round

R150 · Diff two near-identical config blocks

blind round 11/11 local models answered

Every answer below carries the name of the model that wrote it. Blind refers to the scoring: the review was done without knowing which model produced which answer — the published result names them.

Show task

Compare Block A and Block B below and list ALL differences between them (there are exactly 3). For each difference, name the key and both values as "key: A-value -> B-value". Output ONLY the list of differences — no explanation.

Block A:
host: api.example.com
port: 8443
timeout: 30
retries: 3
debug: false
region: eu-west-1
max_connections: 100
protocol: https

Block B:
host: api.example.com
port: 8443
timeout: 45
retries: 3
debug: true
region: eu-west-1
max_connections: 150
protocol: https

The answers

Local models (Brain cluster) — identical prompt, shuffled order (the sequence is not a ranking)

gemma4:26b

timeout: 30 -> 45
debug: false -> true
max_connections: 100 -> 150

qwen3.5:9b

timeout: 30 -> 45
debug: false -> true
max_connections: 100 -> 150

qwen3:14b

timeout: 30 -> 45
debug: false -> true
max_connections: 100 -> 150

mistral-small:24b

```diff
timeout: 30 -> 45
debug: false -> true
max_connections: 100 -> 150
```

deepseek-r1:14b

timeout: 30 -> 45  
debug: false -> true  
max_connections: 100 -> 150

llama3.1:8b

* key: timeout -> A-value: 30 -> B-value: 45
* key: max_connections -> A-value: 100 -> B-value: 150
* key: debug -> A-value: false -> B-value: true

qwen3-coder:30b

timeout: 30 -> 45
debug: false -> true
max_connections: 100 -> 150

qwen3:8b

timeout: 30 -> 45
debug: false -> true
max_connections: 100 -> 150

deepseek-r1:32b

- timeout: A-30 -> B-45
- debug: A-false -> B-true
- max_connections: A-100 -> B-150

command-r:35b

- timeout: A-30 -> B-45
- debug: A-false -> B-true
- max_connections: A-100 -> B-150

qwq:32b

timeout: 30 -> 45  
debug: false -> true  
max_connections: 100 -> 150

Outside the Brain cluster — same task, same character budget, collected through their own tooling (response times not comparable)

Claude Opus 5frontier reference

timeout: 30 -> 45
debug: false -> true
max_connections: 100 -> 150

Claude Sonnet 5frontier reference

timeout: 30 -> 45
debug: false -> true
max_connections: 100 -> 150

Claude Fable 5frontier reference

timeout: 30 -> 45
debug: false -> true
max_connections: 100 -> 150

Claude Haiku 4.5frontier reference

timeout: 30 -> 45
debug: false -> true
max_connections: 100 -> 150

Reference

Gold — the correct answerreference

1. timeout: 30 -> 45. 2. debug: false -> true. 3. max_connections: 100 -> 150. (host, port, retries, region und protocol sind in beiden Bloecken identisch.) Maschinell verifiziert per Zeilenvergleich (8 Zeilen je Block, exakt 3 abweichende Zeilen: 3, 5, 7).

Results at a glance— every model in this round with score, response time and status.

Who wrote what

Model	Blind score	Latency	Status
gemma4:26b	10.0	74.9 s	OK
qwen3.5:9b	10.0	14.0 s	OK
qwen3:14b	10.0	9.4 s	OK
mistral-small:24b	7.0	42.8 s	OK
deepseek-r1:14b	10.0	16.9 s	OK
llama3.1:8b	10.0	32.7 s	OK
qwen3-coder:30b	10.0	9.0 s	OK
qwen3:8b	10.0	9.0 s	OK
deepseek-r1:32b	10.0	36.8 s	OK
command-r:35b	10.0	64.0 s	OK
qwq:32b	10.0	60.9 s	OK

Verdict

10 of 11 answering local models matched the gold answer (blind score ≥ 8): gemma4:26b, llama3.1:8b, qwen3-coder:30b, qwen3:14b, qwen3.5:9b, qwen3:8b, deepseek-r1:14b, deepseek-r1:32b, qwq:32b, command-r:35b.