Structure & Data · blind round

R148 · Table QA: sum, filter, max over order data

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

Answer all 3 questions using ONLY the table below. Output ONLY the 3 numbered answers — no explanation.

| OrderID | Customer | Product | Qty | Price | Status |
|---|---|---|---|---|---|
| 1001 | Mueller | Lamp | 2 | 19.99 | shipped |
| 1002 | Schmidt | Chair | 1 | 89.50 | pending |
| 1003 | Mueller | Lamp | 1 | 19.99 | shipped |
| 1004 | Weber | Desk | 1 | 149.00 | shipped |
| 1005 | Schmidt | Chair | 3 | 89.50 | cancelled |
| 1006 | Mueller | Mug | 5 | 6.50 | shipped |

1. (Sum) What is the total revenue (Qty x Price) of all orders with Status "shipped"?
2. (Filter) List the OrderIDs of all orders placed by customer "Mueller".
3. (Max) Which single order has the highest total value (Qty x Price), and what is that value?

The answers

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

deepseek-r1:14b

1. The total revenue for shipped orders is calculated by summing up the products of Qty and Price for each "shipped" status entry. 

   - Order 1001: 2 * 19.99 = 39.98
   - Order 1003: 1 * 19.99 = 19.99
   - Order 1004: 1 * 149.00 = 149.00
   - Order 1006: 5 * 6.50 = 32.50

   Adding these together: 39.98 + 19.99 + 149.00 + 32.50 = 241.47.

2. To find the OrderIDs for "Mueller" customers, we look at all entries where Customer is Mueller:

   - Order 1001
   - Order 1003
   - Order 1006

3. The order with the highest total value is determined by calculating Qty * Price for each order and identifying the maximum:

   - Order 1001: 2 * 19.99 = 39.98
   - Order 1002: 1 * 89.50 = 89.50
   - Order 1003: 1 * 19.99 = 19.99
   - Order 1004: 1 * 149.00 = 149.00
   - Order 1005: 3 * 89.50 = 268.50
   - Order 1006: 5 * 6.50 = 32.50

   The highest value is 268.50 from Order 1005.

**Answers:**

1. $241.47  
2. 1001, 1003, 1006  
3. OrderID: 1005; Value: $268.50

qwen3-coder:30b

1. 241.47
2. 1001, 1003, 1006
3. Order 1005, $268.50

gemma4:26b

1. 241.47
2. 1001, 1003, 1006
3. 1005, 268.50

qwen3.5:9b

1. 241.47
2. 1001, 1003, 1006
3. 1005 and 268.50

llama3.1:8b

1. $1249.90
2. 1001, 1003, 1006
3. 1006 ($162.50)

mistral-small:24b

1. 405.93
2. 1001, 1003, 1006
3. 1004, 149

qwen3:8b

241.47  
1001, 1003, 1006  
1005, 268.50

qwen3:14b

1. 241.47
2. 1001, 1003, 1006
3. 1005, 268.50

qwq:32b

1. 241.47  
2. 1001, 1003, 1006  
3. 1005, 268.50

deepseek-r1:32b

1. The total revenue from orders with status "shipped" is $241.47.

2. The OrderIDs for customer Mueller are 1001, 1003, and 1006.

3. The highest total value is from order ID 1005 with a value of $268.50.

**Answers:**

1. \$241.47  
2. 1001, 1003, 1006  
3. OrderID: 1005; Value: \$268.50

command-r:35b

1. 245.86
2. [1001, 1003, 1006]
3. 268.50

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

Claude Opus 5frontier reference

1. 241.47 (1001: 39.98 + 1003: 19.99 + 1004: 149.00 + 1006: 32.50)
2. 1001, 1003, 1006
3. Order 1005 — 268.50

Claude Sonnet 5frontier reference

1. 241.47
2. 1001, 1003, 1006
3. Order 1005, value 268.50

Claude Fable 5frontier reference

1. 240.47
2. 1001, 1003, 1006
3. Order 1005 (Schmidt, Chair, 3 x 89.50) with a total value of 268.50

Claude Haiku 4.5frontier reference

1. 241.47
2. 1001, 1003, 1006
3. 1005, 268.50

Reference

Gold — the correct answerreference

1. 241.47 (Summe der shipped-Bestellungen: 1001=2x19.99=39.98, 1003=1x19.99=19.99, 1004=1x149.00=149.00, 1006=5x6.50=32.50 -> 39.98+19.99+149.00+32.50=241.47). 2. 1001, 1003, 1006. 3. Bestellung 1005 mit Gesamtwert 268.50 (3x89.50) — hoechster Wert ueber ALLE Bestellungen, obwohl Status 'cancelled' (bewusste Falle: nicht mit shipped-Filter verwechseln). Alle 3 Werte in Python nachgerechnet (siehe Verifikations-Ausgabe des Autoring-Skripts).

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

Who wrote what

Model	Blind score	Latency	Status
deepseek-r1:14b	10.0	27.6 s	OK
qwen3-coder:30b	10.0	13.7 s	OK
gemma4:26b	10.0	134.8 s	OK
qwen3.5:9b	10.0	30.5 s	OK
llama3.1:8b	4.0	46.4 s	OK
mistral-small:24b	4.0	58.8 s	OK
qwen3:8b	10.0	14.7 s	OK
qwen3:14b	10.0	21.4 s	OK
qwq:32b	10.0	88.2 s	OK
deepseek-r1:32b	10.0	59.9 s	OK
command-r:35b	4.0	85.5 s	OK

Verdict

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