Beating GPT-5.6 Sol on retrieval with 100x cheaper open models
- ID
- 11456
- Status
- summarized
- Published
- 06 Aug 2026, 2:18 AM
- Fetched
- 07 Aug 2026, 9:51 PM
- Provider
- Hacker News
- Category
- dev-community
- Original URL
- https://neon.com/blog/how-castform-neon-beats-frontier-models-on-price-and-efficiency
- Source URL
- https://hnrss.org/best
Summary
- Score
- 7.5
- Created
- 07 Aug 2026, 9:51 PM
- Tags
- Audience
- developersai_ml_learnersai_agent_userssaas_founders
What happened
Castform and Neon demonstrate that a 4B open-source model, post-trained with reinforcement learning on retrieval tasks, matches the accuracy of GPT-5.6 Sol while costing 100x less per request. The approach uses Neon's Lakebase Search extensions for corpus storage and synthetic data generation, then RL-trains the small model to do agentic multi-hop search instead of relying on expensive frontier model API calls.
Why it matters
If you're building agentic RAG or multi-hop search workflows, this is concrete evidence that RL post-training a small open-weights model on your specific retrieval task can replace frontier model calls that cost ~$0.03 and take >10s each. Evaluate whether your retrieval loop is a good candidate for task-specific post-training rather than defaulting to GPT-class APIs for every search iteration.
Discussion angle
Where's the practical threshold between paying for frontier model retrieval vs. investing in RL post-training a small model — how narrow does the task need to be before the 100x cost savings justifies the engineering overhead?