Head-to-head
deepseek-r1:32b vs GPT-5.5 (Codex)
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. Frontier references sit outside the blind field and are deliberately unscored on text.
| Model | Runnable sims | Blind text avg | Failure modes |
|---|---|---|---|
| deepseek-r1:32b | 25% (24/95) | 8.77 (30 scored) | ERR 42 · STAT 7 · BLANK 19 · HANG 2 · DNF 1 |
| GPT-5.5 (Codex) frontier reference | 96% (91/95) | unscored by design | ERR 2 · STAT 1 · BLANK 1 |
Biggest contrasts
Rounds where the two models diverged most — open them and read both answers side by side.
| Round | deepseek-r1:32b | GPT-5.5 (Codex) |
|---|---|---|
| R175 · Noise flowfield painting additive particle trails | ERR | OK |
| R174 · Recursive fractal tree growing branch by branch | ERR | OK |
| R173 · Metaball lava lamp with heating and cooling blobs | ERR | OK |
| R172 · Raindrops sliding down a window pane | HANG | OK |
| R171 · Fireworks bursting into fading sparks | ERR | OK |
→ Interactive round-by-round comparison · → Full leaderboard
Generated 2026-08-04 from the published rounds. Harness failures are excluded from every base — a broken cluster node is our fault, not the model's.