Agentset

[Agentset @agentset_ai]

https://t.co/hEsxGXPZeo

[Agentset @agentset_ai]

Do you want your AI SDK app to answer from your own docs?

We wrote about how that works in the blog post below:

[Agentset @agentset_ai]

https://t.co/UTBXwjiyLq

[Agentset @agentset_ai]

ZeroEntropy recently released zembed-1. We ran it through our embedding leaderboard - it's now #1!

0.946 NDCG@10. 55–80% win rate across 16 models.

Impressive work by @ZeroEntropy_AI!

Full breakdown below.

[Agentset @agentset_ai]

@AiAristotle @perplexity_ai Will do 👍

[Agentset @agentset_ai]

https://t.co/3LOloUwq01

[Agentset @agentset_ai]

We tested Claude Opus 4.6 for RAG.

Key takeaways:

- Best for for factual, doc-based Q&A
- Clear upgrade over 4.5 on harder questions

More context is in the blog:

[Agentset @agentset_ai]

https://t.co/NVYqHqBVsp

[Agentset @agentset_ai]

We tested different ways to detect hallucinations in RAG.

LLM judges, atomic claims, encoder-based NLI.
Each comes with clear trade-offs in accuracy, latency, and cost.

Write-up + benchmarks:

[Agentset @agentset_ai]

https://t.co/jA6vpNQvlD

[Agentset @agentset_ai]

New: RAG prompt templates library.

Constraint-first prompts for grounded answers, citations, uncertainty, and response length control.

Copy, use, upvote/downvote what’s actually useful.

🔗👇

[Agentset @agentset_ai]

https://t.co/eSolhQzgYm

[Agentset @agentset_ai]

To save you some time, we made a list of 𝐚𝐰𝐞𝐬𝐨𝐦𝐞 𝐫𝐞𝐫𝐚𝐧𝐤𝐞𝐫𝐬 - a collection of useful resources on top reranking models (contributions welcome).

Check the link below 🔗

[Agentset @agentset_ai]

https://t.co/0OsdzI7Tga

[Agentset @agentset_ai]

Tested Cohere Rerank 4 on our reranker benchmark.

- Pro: now #2 overall, a big jump from v3.5
- Still <1s latency, ~2× slower than zerank-2
- Great on business and finance-heavy workloads

Read full breakdown in the link below 🔗

[Agentset @agentset_ai]

https://t.co/keN6uV3LVG

[Agentset @agentset_ai]

New benchmark: GPT-5.2 for RAG.

• sits in the top cluster, below GPT-5.1 overall
• ~70% fewer tokens per answer
• best model in our pool on scientific claim verification

Details are in the link below 🔗

[Usman S. @MaxProgramming1]

We at @agentset_ai just released a cookbook for a RAG-powered support bot that actually works.

Tired of customers drowning in 100-page manuals? We show you how to:
• Suck in PDFs w/ smart metadata (product/year stuff)
• Filter searches so oven Qs don't pull washer crap
• Get LLM answers that cite the damn page

Tested on real Panasonic ovens & LG washers. Full Python/TS code included. Scales from 5 to 5k docs.

Steal it here: https://t.co/pIi1nGSOvU

[Agentset @agentset_ai]

https://t.co/HXIjJeWD1p

[Agentset @agentset_ai]

We just shipped our Vector Database Leaderboard. 🚀

Pinecone vs Turbopuffer vs Qdrant vs pgvector (and more) – compared by deployment, cost, and where they actually fit in RAG.

🔗 Link below.