Static-position benchmark used to seed ratings before a model plays its first game
Each model chooses a move in 50 roughly equal positions, scored by Stockfish at depth 30.
Centipawn loss, best-move rate, and legal-move rate predict a game rating:
1298.57 − 200.43·ln(CPL+1) + 15.39·best% + 5.85·survival40.
A game-like panel and continuation-stability probe can only lower that prediction.
57
Current core results
44
With game ratings
220
MAE
276
RMSE
-65
Bias
| # | Player | Predicted ▼ | Game Rating | Error | Eq. CPL | Best% | Legal% | Game-like CPL | Stability Risk |
|---|---|---|---|---|---|---|---|---|---|
| 1 | gemini-4-carbon (medium) | 2410 | 2243 | -167 | 15 | 70% | 100% | 16 | 0.0% |
| 2 |
gemini-4-carbon (high)
Supplements deferred using gemini-4-carbon (medium) |
2174 | 2101 | – | 79 | 76% | 100% | Deferred | Deferred |
| 3 |
gemini-3.1-pro-preview (high)
Supplements deferred using gemini-3.1-pro-preview (medium) |
2170 | 2178 | +8 | 37 | 66% | 100% | Deferred | Deferred |
| 4 | gemini-3-flash-preview (medium) | 2085 | 1866 | -219 | 78 | 70% | 100% | – | – |
| 5 | gemini-3.8-flash (high) | 2050 | 2003 | – | 68 | 66% | 100% | 121 | 0.0% |
| 6 | gemini-4-carbon (low) | 2029 | 2150 | +121 | 76 | 66% | 100% | 65 | 0.0% |
| 7 | gemini-4-argon (low) | 1990 | 1758 | – | 79 | 64% | 100% | 65 | 3.1% |
| 8 | claude-fable-5.2 (high) | 1901 | 1867 | -34 | 78 | 58% | 100% | 12 | 0.0% |
| 9 | gemini-3.6-flash (medium) | 1878 | 1889 | +11 | 75 | 56% | 100% | 75 | 0.0% |
| 10 | gemini-3.5-flash (medium) | 1860capped | 2018 | +158 | 68 | 72% | 100% | 136 | 0.0% |
| 11 | gpt-6-astra (medium) | 1797 | 1864 | +67 | 83 | 52% | 100% | 117 | 0.0% |
| 12 | gemini-3.1-pro-preview (medium) | 1780 | 2037 | +257 | 266 | 66% | 100% | 12 | 0.0% |
| 13 | claude-opus-5.5 (high) | 1766 | 1783 | – | 112 | 54% | 100% | 22 | 0.0% |
| 14 | gpt-3.5-turbo-instruct | 1763 | 1643 | -120 | 134 | 56% | 100% | 70 | 0.0% |
| 15 | gpt-6-astra (high) | 1759 | 1996 | +237 | 86 | 50% | 100% | 56 | 0.0% |
| 16 | gemini-4-argon (high) | 1673capped | 2124 | +451 | 74 | 66% | 100% | 225 | 0.0% |
| 17 | gemini-4-argon (medium) | 1670capped | 2025 | +355 | 80 | 64% | 100% | 128 | 0.0% |
| 18 | gpt-5.6-sol (xhigh) | 1608 | 1495 | -113 | 291 | 56% | 100% | 114 | 0.0% |
| 19 | gpt-5.6-sol (low) | 1464 | 1225 | -239 | 341 | 56% | 98% | 372 | 0.0% |
| 20 | gpt-6.1-sol (high) | 1462 | 1740 | +278 | 186 | 48% | 98% | – | – |
| 21 | gpt-5.6-sol (high) | 1440 | 1587 | +147 | 243 | 50% | 98% | 87 | 3.1% |
| 22 | gpt-5.1 (high) | 1388 | 1430 | +42 | 641 | 52% | 100% | – | – |
| 23 | deepseek-v4.1-flash (high) | 1352 | 1270 | – | 356 | 42% | 100% | 250 | 3.1% |
| 24 | gpt-5.6-sol (medium) | 1351capped | 1507 | +156 | 113 | 52% | 100% | 312 | 0.0% |
| 25 | gpt-5.6-luna (xhigh) | 1344 | 1585 | +241 | 393 | 50% | 98% | 244 | 3.1% |
| 26 | gemini-3.8-flash (medium) | 1303 | 1453 | – | 370 | 66% | 94% | 208 | 0.0% |
| 27 | gpt-5.6-terra (xhigh) | 1284capped | 1250 | -34 | 225 | 48% | 100% | 340 | 0.0% |
| 28 | deepseek-v4-flash-0731 (max) | 1266capped | 1417 | +151 | 146 | 38% | 100% | 229 | 3.1% |
| 29 | deepseek-v4-flash-0731 (high) | 1245 | 1584 | +339 | 256 | 38% | 98% | 316 | 0.0% |
| 30 | claude-fable-5.1 (high) | 1190 | 1164 | – | 534 | 44% | 98% | 217 | 0.0% |
| 31 | gpt-5.6-luna (medium) | 1152 | 307 | -845 | 555 | 42% | 98% | 383 | 0.0% |
| 32 | gpt-5.6-luna (high) | 1133 | 735 | -398 | 610 | 42% | 98% | 190 | 3.2% |
| 33 | gpt-5.5 (low) | 1104 | 523 | -581 | 561 | 50% | 96% | – | – |
| 34 | gpt-5.6-terra (low) | 983 | 472 | -511 | 475 | 40% | 96% | 303 | 3.1% |
| 35 | gpt-5.2 (high) | 928capped | 1072 | +144 | 650 | 50% | 100% | 231 | – |
| 36 | gpt-5.5 (xhigh) | 897 | 501 | -396 | 853 | 42% | 96% | 157 | 0.0% |
| 37 | gpt-6.1-sol (medium) | 859 | 1390 | – | 488 | 46% | 92% | – | – |
| 38 | gpt-5.6-terra (high) | 831capped | 506 | -325 | 617 | 52% | 94% | 438 | 3.1% |
| 39 | deepseek-v4-flash (max) | 790 | 920 | +130 | 579 | 30% | 96% | 380 | 3.1% |
| 40 | gpt-5.1 (no thinking) | 718 | 487 | -231 | 799 | 38% | 94% | 429 | 9.4% |
| 41 | gpt-5.6-luna (low) | 667 | 457 | -210 | 637 | 40% | 90% | 551 | 12.5% |
| 42 | gemini-2.5-flash (no thinking) | 546capped | 361 | -185 | 639 | 42% | 92% | 744 | 16.1% |
| 43 | gpt-5.6-terra (medium) | 517 | 567 | +50 | 877 | 36% | 88% | 352 | 6.5% |
| 44 | gemini-3.1-flash-lite-preview | 500capped | 193 | -307 | 409 | 58% | 96% | 483 | 25.0% |
| 45 | gpt-5.1-chat | 489 | 546 | +57 | 1052 | 32% | 92% | 795 | 19.4% |
| 46 | gemma-4-31b-it (high) | 303 | 159 | – | 1065 | 26% | 82% | 810 | 28.7% |
| 47 | gpt-oss-120b (low) | 181 | 188 | +7 | 1142 | 16% | 90% | 860 | 6.2% |
| 48 | gemma-4-31b-it (no thinking) | 122 | -217 | – | 1920 | 22% | 72% | 1547 | 67.8% |
| 49 | kimi-k2-0905 | 114 | -193 | -307 | 1998 | 22% | 68% | 1439 | 42.4% |
| 50 | llama-4-maverick | 56 | -171 | -227 | 1962 | 18% | 68% | 1582 | 54.3% |
| 51 | qwen3-235b-a22b-2507 | 24 | -368 | -392 | 1695 | 14% | 70% | 1420 | 80.9% |
| 52 | deepseek-v4-pro (no thinking) | -23 | -102 | – | 2135 | 14% | 64% | 1560 | 46.6% |
| 53 | deepseek-v4-flash (no thinking) | -51 | -81 | – | 2871 | 16% | 46% | 1900 | 81.7% |
| 54 | glm-4.6 (no thinking) | -120capped | -296 | -176 | 1930 | 18% | 68% | 2242 | 80.3% |
| 55 | deepseek-v4-flash-0731 (no thinking) | -183 | -199 | – | 3005 | 8% | 42% | 2146 | 95.8% |
| 56 | mistral-medium-3 | -218 | -275 | -57 | 3069 | 6% | 42% | 2971 | 75.0% |
| 57 | gpt-3.5-turbo | -299 | -500 | -201 | 3381 | 2% | 38% | 3914 | 112.5% |
| 58 | eubos Engine Legacy | 2197 | 2357 | – | 20 | 60% | 100% | – | – |
| 59 | gemini-3-flash-preview (high) Legacy | 2071 | 1755 | – | 61 | 66% | 100% | – | – |
| 60 | maia-1900 Engine Legacy | 1955 | 1816 | – | 81 | 62% | 100% | – | – |
| 61 | gemini-3-pro-preview (high) Legacy | 1658 | 1963 | – | 488 | 66% | 100% | – | – |
| 62 | maia-1100 Engine Legacy | 1627 | 1628 | – | 264 | 56% | 100% | – | – |
| 63 | survival-bot Engine Legacy | 1593 | 1669 | – | 78 | 38% | 100% | – | – |
| 64 | grok-4.1-fast Legacy | 1405 | 1445 | – | 233 | 40% | 100% | – | – |
| 65 | gpt-5 (medium) Legacy | 1162 | 1085 | – | 366 | 30% | 100% | – | – |
| 66 | grok-4.3 (medium) Legacy | 1104 | 936 | – | 444 | 36% | 98% | – | – |
| 67 | gpt-oss-120b (high) Legacy | 1056 | 681 | – | 335 | 22% | 100% | – | – |
| 68 | grok-4-fast Legacy | 1036 | 1269 | – | 394 | 30% | 98% | – | – |
| 69 | gpt-5.5 (medium) Legacy | 1034 | 464 | – | 502 | 44% | 96% | – | – |
| 70 | gpt-5.4 (high) Legacy | 950 | 1148 | – | 774 | 26% | 100% | – | – |
| 71 | gemini-3-pro-preview (medium) Legacy | 925 | 623 | – | 615 | 48% | 94% | – | – |
| 72 | gpt-5.4 (minimal) Legacy | 900 | 667 | – | 665 | 28% | 98% | – | – |
| 73 | gpt-oss-120b (medium) Legacy | 756 | 516 | – | 738 | 20% | 98% | – | – |
| 74 | claude-opus-4.6 (medium) Legacy | 717 | 1174 | – | 434 | 30% | 94% | – | – |
| 75 | grok-4.3 (high) Legacy | 696 | 1081 | – | 584 | 24% | 96% | – | – |
| 76 | gemini-2.5-pro (default) Legacy | 620 | 345 | – | 806 | 40% | 90% | – | – |
| 77 | deepseek-v3.2 (thinking) Legacy | 620 | -55 | – | 920 | 14% | 98% | – | – |
| 78 | random-bot Engine Legacy | 593 | 400 | – | 990 | 6% | 100% | – | – |
| 79 | gpt-5.2 (no thinking) Legacy | 549 | 645 | – | 1006 | 30% | 94% | – | – |
| 80 | gpt-5.5 (no thinking) Legacy | 487 | 103 | – | 823 | 34% | 86% | – | – |
| 81 | gpt-5.5 (high) Legacy | 438 | 378 | – | 1116 | 34% | 88% | – | – |
| 82 | claude-opus-4.6 (no thinking) Legacy | 389 | 7 | – | 807 | 22% | 92% | – | – |
| 83 | claude-opus-4.5 (high) Legacy | 347 | 1010 | – | 1109 | 28% | 88% | – | – |
| 84 | claude-opus-4.7 (no thinking) Legacy | 344 | -48 | – | 1400 | 32% | 84% | – | – |
| 85 | deepseek-v4-pro (max) Legacy | 330 | 310 | – | 1286 | 30% | 84% | – | – |
| 86 | deepseek-v4-pro (high) Legacy | 328 | -48 | – | 1299 | 30% | 84% | – | – |
| 87 | mimo-v2-flash (medium) Legacy | 306 | 235 | – | 991 | 14% | 94% | – | – |
| 88 | deepseek-v4-flash (high) Legacy | 269 | 225 | – | 1076 | 18% | 92% | – | – |
| 89 | gemini-2.0-flash-001 Legacy | 176 | 253 | – | 1359 | 18% | 90% | – | – |
| 90 | gpt-5-chat Legacy | 167 | 256 | – | 1346 | 20% | 84% | – | – |
| 91 | gpt-5.4 (no thinking) Legacy | 108 | 172 | – | 1123 | 14% | 82% | – | – |
| 92 | kimi-k2.5 (no thinking) Legacy | -111 | -202 | – | 2441 | 10% | 62% | – | – |
| 93 | grok-3-mini Legacy | -175 | 44 | – | 2475 | 6% | 66% | – | – |
| 94 | deepseek-chat-v3-0324 Legacy | -216 | -181 | – | 3038 | 6% | 48% | – | – |
| 95 | glm-4.6 (thinking) Legacy | -246 | -226 | – | 3023 | 4% | 46% | – | – |
| 96 | deepseek-v3.2 (no thinking) Legacy | -258 | -166 | – | 3730 | 6% | 36% | – | – |
| 97 | kimi-k2 Legacy | -265 | -138 | – | 2855 | 2% | 56% | – | – |
| 98 | gpt-3.5-turbo-0613 Legacy | -272 | -250 | – | 2530 | 0% | 54% | – | – |
| 99 | deepseek-v3.1-terminus (no thinking) Legacy | -326 | -242 | – | 3861 | 2% | 34% | – | – |
| 100 | deepseek-r1-distill-qwen-32b Legacy | -339 | -265 | – | 4127 | 2% | 24% | – | – |
| 101 | deepseek-chat-v3.1 (no thinking) Legacy | -354 | -324 | – | 4431 | 2% | 22% | – | – |
| 102 | mimo-v2-flash (no thinking) Legacy | -384 | -500 | – | 4416 | 0% | 22% | – | – |
| 103 | llama-3.3-70b-instruct Legacy | -415 | -500 | – | 5153 | 0% | 2% | – | – |
Predicted = Rating predicted from position results; capped means the game-like panel or stability probe lowered it
Error = Game rating minus prediction, shown once the game rating is established (RD ≤ 150); summary stats cover current results only
Eq. CPL = Average centipawn loss on 50 equal positions (illegal moves score as a large loss)
Best% = Moves matching Stockfish's best move; Legal% = Legal moves on the first attempt
Stability Risk = Forfeit rate plus catastrophic-blunder rate when play continues from probe positions
Deferred means a lower reasoning effort qualified this player to skip both supplemental panels. Each prediction uses the player's own measurements.
Legacy = Results from an older protocol; shown for reference but not used to seed ratings