Head-to-head
deepseek-r1:14b vs deepseek-r1:32b
Both models answered the same one-shot tasks under identical rules — no retries, answers published unedited. Two separate axes, never blended: does the code actually run, and how do blind-scored text answers hold up against a written gold answer.
| Model | Runnable sims | Blind text avg | Failure modes |
|---|---|---|---|
| deepseek-r1:14b | 14% (12/85) | 8.09 (32 scored) | ERR 55 · STAT 4 · BLANK 10 · HANG 1 · DNF 3 |
| deepseek-r1:32b | 27% (23/85) | 8.90 (20 scored) | ERR 34 · STAT 7 · BLANK 19 · HANG 1 · DNF 1 |
Biggest contrasts
Rounds where the two models diverged most — open them and read both answers side by side.
| Round | deepseek-r1:14b | deepseek-r1:32b |
|---|---|---|
| R155 · Linked list with animated pointer relinking | BLANK | OK |
| R152 · Hash table with visible collision chains | OK | BLANK |
| R140 · Water sloshing in a tilting container | ERR | OK |
| R136 · Double pendulum tracing a chaotic path | ERR | OK |
| R123 · Breadth-first flood fill spreading through a region | ERR | OK |
→ Interactive round-by-round comparison · → Full leaderboard
Generated 2026-07-31 from the published rounds. Harness failures are excluded from every base — a broken cluster node is our fault, not the model's.