A new category of "decision models" has emerged, with multiple companies releasing products that provide typed answers like probabilities or scores instead of free text. Jev is positioned as the benchmark for this category, with competitors like OpenAI's Decisions API, Microsoft-Decision-1, and Perplexity's pplx-decider-v1.1-27b offering various features and pricing. These models are being integrated into serving and routing systems, and tools are available for users to train their own decision models using smaller, open-source models. The development of these decision models is also being explored in academic research, with a focus on improving reasoning latency within agents. AI
IMPACT Accelerates development of AI agents by providing structured outputs and improving cost-efficiency for routing tasks.
RANK_REASON Multiple companies releasing new product categories and benchmarks, indicating a significant shift in the AI product landscape. [lever_c_demoted from significant: ic=1 ai=1.0]
- Anthropic
- Apple Inc.
- Cloudflare
- CMU
- GPT-6 Luna
- Hugging Face
- Jevíčko
- LangChain
- Microsoft
- OpenAI
- Opus 5.5
- Perplexity
- Qwen3.5 4B
- Vercel
- vLLM
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
