decision model · League 1

Perplexity Decider 27B v1.1NVIDIA RTX PRO 6000 Blackwell + NVIDIA DGX Spark

FinishedPublished

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

How we called it

Model ID
pplx-decider-v1.1-27b
Provider or endpoint
NVIDIA RTX PRO 6000 Blackwell (96 GB), bf16; NVIDIA DGX Spark (GB10) from 16 Oct
Date run
2026-10-15 to 2026-10-26
Price
no price per token; estimate: GPU time, at most; source content/own_cost.json (wall clock on our machines × hourly rate); price date not recorded
Model type
decision model
Weights
open
Runs on
our GPUs
Reasoning setting
none (decision model)
Confidence source
decision-model probability
Rules text
long
Option groups
up to 25 options per request; more legal plays are split into equal groups, a winner per group, then among the winners; in this data 226 decisions were split into groups of 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25
Sampling settings
not sent (decision request)
Notes
Perplexity's released checkpoint on our GPUs; moved to a DGX Spark on 16 Oct 11:04 UTC (same checkpoint and server arguments; 20 replayed requests: same play 20/20)

Missing B7 fields: price.date

Results in League 1

Error rate (PR)
38.835.3 to 42.5
Decisions
2,052 (2,052 scored, 0 invalid)
Engine's best play
47.1%
Blunder rate
26.8%
P95 equity loss
0.389
Mean loss, worst 5%
0.569
Median / P90 s per decision
1.97 / 7.48
$ per 1,000 decisions
data-v="6.593">6.593estimate · RTX PRO 6000 Blackwell
Risk judgement, margin 0.001
56.9 (95% interval 54.9 to 58.8); risks found 33.1% (233 / 703), right safety 80.7% (682 / 845)
Risk judgement, margin 0.02
56.7 (95% interval 54.7 to 58.7); risks found 33.7% (211 / 626), right safety 79.7% (735 / 922)
No trade-off
504 (left out of every risk metric)

JSON

Risk call card

56.9risk judgement, margin 0.001

right risks233
wrong risks67
needless risks163
missed risks403
right safety682

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
2,052
Eligible (full probability distribution)
1,633
Left out, by reason
asked in groups of options 202; not audited 217
Close choices
40.0% (654 / 1,633)
Ties in the top two
80 decisions (80 at a positive probability, 0 at 0; the two most likely plays tied: 40, the second tied with the third: 40); 56 of them close choices. Ties are broken by the order the plays were shown, never by the engine.
Between a safe play and a riskier one
46.2% (302 / 654) of the close choices; 18.5% (302 / 1,633) of all eligible decisions
Probabilities of the two most likely plays, mean with median
DecisionsNumberp1 (most likely play)p2 (second)p1 + p2
Close choices6540.391median 0.3850.279median 0.2700.670median 0.665
Other decisions9790.717median 0.7290.134median 0.1310.850median 0.892
How much probability the two plays hold together: decisions per bin
p1 + p20.0–0.10.1–0.20.2–0.30.3–0.40.4–0.50.5–0.60.6–0.70.7–0.80.8–0.90.9–1.0
Close choices00429801211501207278
Other decisions0002195398137202468
What differs between the two most likely plays (several can differ; none is picked as the cause)
DecisionsNumberHitting rollsNewly made pointsPip countsLongest primeNone / 1 / 2 / 3 / 4 differ
Close choices65454.4%356 / 65426.8%175 / 65414.8%97 / 65428.3%185 / 654160 / 262 / 154 / 69 / 9
Other decisions97954.3%532 / 97933.9%332 / 97912.4%121 / 97930.8%302 / 979205 / 394 / 259 / 109 / 12
Engine values (GNUbg 2-ply, equity), mean with median
DecisionsNumberGap between the two plays' valuesMost likely play: behind the engine's bestSecond most likely play: behind the engine's best
Close choices6540.134median 0.0720.101median 0.0340.110median 0.049
Other decisions9790.143median 0.0790.053median 0.0000.140median 0.065
Close choices between a safe play and a riskier one, by position class
MarginIn Risk pays positionsIn Safety pays positions
margin 0.00166.4%170 / 25650.2%132 / 263
margin 0.0266.5%151 / 22751.7%151 / 292
Close choices among eligible decisions, by position class and stake band
MarginRisk paysSafety paysNo trade-offRisk pays, small stakeRisk pays, medium stakeRisk pays, big stake
margin 0.00148.5%256 / 52839.0%263 / 67431.3%135 / 43150.9%29 / 57, under 200 decisions44.3%74 / 167, under 200 decisions50.3%153 / 304
margin 0.0248.2%227 / 47139.9%292 / 73131.3%135 / 431N/A0 / 044.3%74 / 167, under 200 decisions50.3%153 / 304

At the 0.02 margin a Risk pays position has a stake of at least 0.02, so the small band is empty. A big stake means playing safe costs more by the engine's value, not that the decision was harder.

Missed risks where the risk is worth taking, close choices vs other decisions
MarginClose choices: missed risksOther decisions: missed risks (weighted to the close choices' mix)Difference, close minus otherCrude difference
margin 0.001365 decisions58.3%crude 57.1% (97 / 170), under 200 decisions76.2%crude 75.4% (147 / 195), under 200 decisions−17.9 points95% interval −27.0 to −5.9−18.3 points95% interval −26.9 to −9.7
margin 0.02326 decisions57.6%crude 56.3% (85 / 151), under 200 decisions76.7%crude 75.4% (132 / 175), under 200 decisions−19.0 points95% interval −29.0 to −6.3−19.1 points95% interval −28.8 to −9.3

By game phase

PhaseError rate (PR)Decisions
Opening38.0304
Contact46.11,277
Holding gameToo few decisions75
Race20.8251
Bear-offToo few decisions145