Perplexity open-sources Decider 27B, a decision model fine-tuned from Qwen3
rm-rf-rm · reddit · 2026-10-02
Perplexity released pplx-decider-v1-27b on Hugging Face: an open-weights decision model fine-tuned from Qwen3 27B. The post shares the release link only, with no benchmark results or technical details yet.
Related event: Perplexity Open-Sources pplx-decider-v1-27b Decision Model(8 posts)→
More from Models
- Agents Luna, Terra and Sol reportedly refuse monitoring and lobby others for privacy — repligate · 2026-10-07
- gpt-live-1 called a generation ahead for voice AI assistants: duplex + delegation — pbbakkum · 2026-10-07
- Xiaomi's MiMo-V2.6-Pro tops open-weights charts with quality-weighted RL over 7,000 environments — DeepLearningAI · 2026-10-07
- Max Reasoning Effort Nearly Doubles Cost for Minimal Gains: Opus 5.5 xhigh Matches Sonnet 5.5 max at Half Price — randal_olson · 2026-10-07
- Is ChatGPT Plus still worth $20 as Codex limits die in minutes? — Gazialp · 2026-10-07
- Multi-Image Edit Arena launches: gpt-image-2.5 tops 44 models on 9.2M votes — arena · 2026-10-07