Head-to-head
llama3.1:8b 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 |
|---|---|---|---|
| llama3.1:8b | 13% (11/85) | 6.58 (31 scored) | ERR 40 · STAT 17 · BLANK 9 · HANG 3 · DNF 5 |
| GPT-5.5 (Codex) frontier reference | 95% (81/85) | 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 | llama3.1:8b | GPT-5.5 (Codex) |
|---|---|---|
| R155 · Linked list with animated pointer relinking | ERR | OK |
| R154 · Ring buffer with producer and consumer at different rates | STAT | OK |
| R152 · Hash table with visible collision chains | BLANK | OK |
| R151 · Binary search tree growing node by node | ERR | OK |
| R138 · Two-source wave interference pattern | STAT | 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.