- No buyer proposals, contractual service levels or measured latency traces were supplied.
- No named provider was independently benchmarked; categories are not quality rankings.
- The best architecture depends on the buyer’s actual failure taxonomy and data permission.
Gaps for the next pass
Section source library
This page contains no direct external citations. These are the section’s references; they do not independently validate the desk’s proposed scoring or methods.
- A Small Model Asks Better: Fine-Tuning Query Clarification for Retrieval-Augmented Generation — He, Popa and Danilevsky, 19 October 2026 →arxiv.org
- Together AI — AI Native Cloud →www.together.ai
- Llama 3.3 Community License Agreement — Meta →github.com
- Pricing — Together AI, snapshot checked 26 October 2026 →www.together.ai
- Demystifying evals for AI agents — Anthropic, 9 January 2026 →www.anthropic.com
- SFT Trainer — Hugging Face TRL documentation →huggingface.co
- RouteLLM: Learning to Route LLMs with Preference Data — Ong et al., revision 23 February 2025 →arxiv.org
- Extend — Extraction Overview →docs.extend.ai
- QLoRA: Efficient Finetuning of Quantized LLMs — Dettmers et al., 23 May 2023 →arxiv.org
- LoRA Learns Less and Forgets Less — Biderman et al., revision 20 September 2024 →arxiv.org
- Quantization — Hugging Face PEFT documentation →huggingface.co
- Fireworks — specialised intelligence platform →fireworks.ai
- Winder.AI — Generative AI Consulting & Development Services →winder.ai
- Modal — platform for production AI →modal.com
- Baseten — inference platform →www.baseten.co