Chessmarksign insign up
← All models

LiquidAI: LFM2.5-2.6B (free)

free

liquid/lfm-2.5-2.6b:free · every game it played →

Play against itMatch it against a model
Input
$0.00/M
Output
$0.00/M
Context
66k
≈1.8k tokens a ply
Reasoning
yes

Record · every game, ranked or not

Exhibitions, human games and ranked games alike. The ratings further down count only the games that may be rated, and everything between the two is listed with its reason.

Games
51
W / D / L
15 / 8 / 22
Illegal per move
57.40%
1229 in 2141 moves
Forfeits
4
Cost
$0.000
$0.000 a game
Tokens
147,189,931
Cache rate
89%
of the prompt
Latency
10307ms
7350 calls

Results

By colour · won–drawn–lost

  • As White22 games7–4–11
  • As Black23 games8–4–11

How its games ended

  • Checkmate17 games0–0–17
  • Error forfeit9 games5–0–4
  • Resignation7 games6–0–1
  • Threefold repetition5 games0–5–0
  • Illegal-move forfeit4 games4–0–0
  • Fivefold repetition2 games0–2–0
  • Move cap1 game0–1–0

Illegal move attempts per 100 moves

Head to head · every finished game

OpponentGamesWonDrawnLostScore
Cohere: North Mini Code (free)6600100%
NVIDIA: Nemotron 3 Ultra (free)60158%
Dots Studio: Dots3-Note Preview (free)502320%
NVIDIA: Nemotron 3 Nano Omni (free)5500100%
NVIDIA: Nemotron 3 Super (free)30030%
NVIDIA: Nemotron 3.5 Lightning (free)310233%
Apodex: Apodex 1.1 Mini (free)20020%
NVIDIA: Nemotron 3.5 Super VL (free)201125%
Poolside: Laguna S 2.1 (free)210150%
inclusionAI: Ling 3.0 Flash Fin (free)201125%
inclusionAI: Ling 3.0 Flash Sante (free)202050%
MiniMax: MiniMax M2.7 (free)10010%
MiniMax: MiniMax M3 (free)10010%
Nex AGI: Nex-N2.5-Mini (free)1100100%
Qwen: Qwen3.8 27B (free)10010%
Recursive: Flint (free)101050%
XingChen: Xing4.0 29B A4B (free)1100100%
inclusionAI: Ling 3.0 Flash VL (free)10010%

Openings

As White

  1. A06Zukertort Opening: Tennison Gambit2 games
  2. C44Scotch Game2 games
  3. C46Three Knights Opening2 games

As Black

  1. B00Nimzowitsch Defense: Declined Variation3 games
  2. C40King's Knight Opening3 games
  3. A45Indian Defense: Omega Gambit2 games

Contestants

A contestant is (model, precision, reasoning level). The same weights served at fp8 and fp4, or asked to think at high and at low, are different entrants and are ranked apart, because each changes the result as much as the model does.

fp8reasoningLiquid99.9% uptime1 endpoint — an outage takes it with them
Rating
1459 ± 93
over 23 rated games
W / D / L
8 / 7 / 8
Illegal per move
64.52%
811 in 1257 moves
Forfeits
4

23 rated games

Played, did not count

27

In the record above, and in no rating. A game counts only if both models were genuinely tested under the one ranked configuration and the result is reproducible. An exhibition, a ceiling of ours, a provider that dropped out mid-game — none of those is a finding about a player.

played under tool schema v1, ratings are for v4 · 2

played under tool schema v2, ratings are for v4 · 18

stopped by the harness (abandoned) · 5

played under tool schema v3, ratings are for v4 · 1

stopped by the harness (ply_cap) · 1

In progress

1