chat model, one letter · League 3

Gemini 4 Argonlow

FinishedPublished

Results f24d1f41, updated 16 November 2026, 21:24 UTC; scored by GNU Backgammon 1.08, 2-ply.

How we called it

Model ID
gemini-4-argon-low
Provider or endpoint
Google Antigravity subscription
Date run
2026-10-20 to 2026-10-24
Price
$2.0 / $10.0 per million tokens in / out; API-equivalent (subscription); price date not recorded
Model type
chat model, one letter
Weights
closed
Runs on
subscription
Reasoning setting
thinking level low, the lowest
Confidence source
none
Rules text
long
Option groups
up to 26 options per request (letters A-Z); more legal plays are split into equal groups, a winner per group, then among the winners; in this data 204 decisions were split into groups of 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26
Sampling settings
temperature 1.0
Notes
Antigravity model id gemini-4-argon-low (model version gemini-4-argon); list price $2/$10 per million tokens until 2026-12-01 ($4/$20 after)
Earlier name
Gemini 4 Argon (off) (renamed on 28 Oct 2026 to the reasoning effort its requests sent)
Reasoning, from the logs
The API reported 0 thinking tokens, but the responses carry thought signatures.

Missing B7 fields: price.date

Results in League 3

Error rate (PR)
16.513.7 to 19.5
Decisions
1,484 (1,484 scored, 0 invalid)
Engine's best play
60.3%
Blunder rate
12.4%
P95 equity loss
0.181
Mean loss, worst 5%
0.331
Median / P90 s per decision
3.82 / 12.2
$ per 1,000 decisions
10.204API-equivalent (subscription)
Risk judgement, margin 0.001
69.2 (95% interval 66.2 to 72.3); risks found 47.8% (209 / 437), right safety 90.7% (543 / 599)
Risk judgement, margin 0.02
68.7 (95% interval 65.6 to 71.9); risks found 49.6% (188 / 379), right safety 87.8% (577 / 657)
No trade-off
448 (left out of every risk metric)

JSON

Risk call card

69.2risk judgement, margin 0.001

right risks209
wrong risks27
needless risks56
missed risks201
right safety543

Hit risk only: being hit. Gammon risk and priming are not in it.

Close choices

A decision is a close choice when the second most likely play has at least half the probability of the first (internally: split).

Probabilities over plays, not sampled reasoning. Method

Decisions
1,484
Eligible (full probability distribution)
0
Left out, by reason
one letter only, or a program 1,484
Close choices
N/A no full probability distribution

By game phase

PhaseError rate (PR)Decisions
Opening10.4208
Contact23.4856
Holding gameToo few decisions80
RaceToo few decisions184
Bear-offToo few decisions156