Advertise here
#424

Decider V1.2

Perplexity Release: 2026-10-17 Tested on: 2026-10-17 10:59 perplexity/pplx-decider-v1.2::none
27BDenseOpen source

Summary

Decider V1.2 scores 2.6 on AI BENCHY and ranks #424. It has 9.4 reliability, a 24.6% pass rate, $0.001 total cost, and 517ms average response time.

Model facts

Researched on 2026-10-17

Reported
Parameters
27B
Architecture
Dense
Availability
Open source
License
Apache-2.0

The official v1.2 model card identifies the Qwen3.8-27B dense backbone, Apache-2.0 weights, a decision readout and the public inference implementation. 27B is nominal; the tensor summary rounds to 26B. Joint requests share state and one sequence across questions. OpenRouter returns typed decisions without generated text or exposed reasoning controls, with up to 255 choice options. Perplexity endpoint policy publishes training=false and retainsPrompts=false. Fallback date records this onboarding request date.

Parameter source Availability source

Score

2.6

Consistency

4.9

Reliability

9.4

Total Cost (Current Price)

$0.001

Total Output Tokens

67

Total Input Tokens

48,363

Input Price

$0.020 / 1M

Output Price

$0.000 / 1M

Cache Read Price

N/A

Cache Write Price

N/A

Cache prices apply to input tokens. Reads reuse cached prompts; writes store them and can cost extra. Output tokens use the output price.

Tests Correct

Wrong Tests: 7

Attempt pass rate: 24.6%

Benchmark coverage: 36/69 attempts. Choices supplied by adapters. Supports 12/23 tests. Other tests are unsupported, not failed. Score uses the full suite.

Flaky tests

1

Flaky tests had mixed outcomes across runs (at least one pass and one fail).

Response Time (avg)

517ms

Response Time (max): 776ms

Response Time (total): 6.20s

Wrong answer: 5 API error: 1 Did not follow instructions: 1

Charts

Choose the first model, then click a second model to open a side-by-side page.

Total Output Tokens

Score vs Total Output Tokens

Quick Compare

Decider V1.2defaultvsJev 1.13defaultDecider V1.2defaultvsDecider V1.1 27BdefaultDecider V1.2defaultvsd1-ultradefaultDecider V1.2defaultvsDrex v1.5defaultDecider V1.2defaultvsSpan-02defaultDecider V1.2defaultvsGPT-6 Luna DecisionsdefaultDecider V1.2defaultvsCommand A+lowDecider V1.2defaultvsKev 4BdefaultDecider V1.2defaultvsCommand A+noneDecider V1.2defaultvsJev 1.14default

Category Breakdown

Category Score Consistency Tests Correct
Agentic 0.0 0.0
Anti-AI Tricks 5.8 7.5
Coding 4.5 6.7
Combined 0.0 0.0
Data parsing and extraction 6.5 10.0
Domain specific 5.9 7.2
General Intelligence 0.0 0.0
Instructions following 1.5 5.0
Puzzle Solving 1.7 3.3
Tool Calling 0.0 0.0
Trivia 0.0 0.0

Compared models