> Canonical page: https://posttrainllm.com/benchmarks

Benchmark archive

# Every score should have a replay.

Protocol, raw decisions, local command, and the specialist beside its larger opponent. Successful and failed attempts both belong here when their evidence can be inspected and replayed.

## Available now

 3 evidence surfaces · 0 qualified ratings

[± Cross-game Arena candidate · ratings unqualified Head-to-head Arena Elo and paired-score evidence, with uncertainty and qualification shown before rank. extensible adapters 95% intervals forfeit-aware no FIDE claim Chess sample 4 games Provisional gap 170 Arena Elo Qualified no Enter →](https://posttrainllm.com/benchmarks/arena)[♞ Character Chess development gate passed Can chess specialization compress frontier move selection into a 30–50M Mac-local model? FEN + UCI 20 tactics full games no tools Frontier 65% exact Strongest local 9B · 10% Our SLM not trained Watch →](https://posttrainllm.com/benchmarks/chess)[2048 Character 2048 failed · advantage not reproduced Can a 30–50M specialist compress game intelligence from a frontier general LLM? character input strict + constrained no tools full replays Best general LLM Qwen3 4B Instruct Mean score 845 Our SLM not trained Watch →](https://posttrainllm.com/benchmarks/game-2048)

 The admission rule

A game is admitted only when a pinned frontier model clearly beats random legal play and the strongest measured local general model. Chess passed that development screen; Character 2048 failed and remains visible as negative evidence.