ArliAI: QwQ 32B RpR v1

Text input Text output Free Option
Author's Description

QwQ-32B-ArliAI-RpR-v1 is a 32B parameter model fine-tuned from Qwen/QwQ-32B using a curated creative writing and roleplay dataset originally developed for the RPMax series. It is designed to maintain coherence and reasoning across long multi-turn conversations by introducing explicit reasoning steps per dialogue turn, generated and refined using the base model itself. The model was trained using RS-QLORA+ on 8K sequence lengths and supports up to 128K context windows (with practical performance around 32K). It is optimized for creative roleplay and dialogue generation, with an emphasis on minimizing cross-context repetition while preserving stylistic diversity.

Key Specifications
Cost
$$
Context
32K
Parameters
32B
Released
Apr 13, 2025
Speed
Ability
Reliability
Supported Parameters

This model supports the following parameters:

Reasoning Include Reasoning Seed Top P Temperature Top Logprobs Logit Bias Logprobs Stop Min P Max Tokens Frequency Penalty Presence Penalty
Features

This model supports the following features:

Reasoning
Performance Summary

ArliAI: QwQ 32B RpR v1, a 32B parameter model fine-tuned for creative writing and roleplay, demonstrates exceptional speed, consistently ranking among the fastest models across eight benchmarks. It also offers highly competitive pricing, placing in the 90th percentile across seven benchmarks. The model exhibits outstanding reliability with a 99% success rate, indicating minimal technical failures. In terms of performance across categories, the model shows strong capabilities in General Knowledge (98.5% accuracy, 63rd percentile), notably being the most accurate model at its price point. Its Reasoning abilities are also impressive (92.0% accuracy, 83rd percentile), and it performs well in Coding (90.0% accuracy, 74th percentile) and Ethics (99.0% accuracy, 64th percentile). However, the model struggles significantly with Hallucinations (72.0% accuracy, 18th percentile), frequently failing to acknowledge uncertainty. Its Instruction Following performance is inconsistent, with one benchmark showing 47.5% accuracy and another indicating 0% accuracy, suggesting a potential area for improvement or an anomaly in the latter test. Email Classification is also a notable weakness (74.0% accuracy, 7th percentile).

Model Pricing

Current Pricing

Feature Price (per 1M tokens)
Prompt $0.02
Completion $0.07

Price History

Available Endpoints
Provider Endpoint Name Context Length Pricing (Input) Pricing (Output)
Chutes
Chutes | arliai/qwq-32b-arliai-rpr-v1 32K $0.02 / 1M tokens $0.07 / 1M tokens
Chutes
Chutes | arliai/qwq-32b-arliai-rpr-v1 32K $0.02 / 1M tokens $0.07 / 1M tokens
Benchmark Results
Benchmark Category Reasoning Strategy Free Executions Accuracy Cost Duration