Not Diamond Free

-
Developer Not Diamond, Inc.
Region United States
Official Website https://www.notdiamond.ai/

Not Diamond provides multi-model routing capabilities to help teams dynamically select models between quality, latency, and cost.

Not Diamond Product Interface

Not Diamond tool text

Core parameters and statistics

Parameters Current public information Description
Product positioning Multi-model routing layer Dynamic balance between effect, cost and delay
Delivery form API + console Adapt to existing LLM application stack
Key capabilities Routing, evaluation, policy control Reduce the cost of manual model cutting
Target users AI application team, platform engineering team Emphasis on contextual controllability of production
Official Documentation Docs Site https://docs.notdiamond.ai/

The essence of Not Diamond is the "decision-making layer" rather than the "model layer". Its value lies in turning model selection from manual experience to configurable strategies.

User and market recognition

  • The official website FAQ publicly claims: The team using Not Diamond achieved more than 30%+ cost savings under the coding agent workload (official claim, non-third-party independent audit conclusion).
  • The product is positioned as a routing layer, not a gateway - the actual model request is still executed on the client through its own gateway, and the routing layer is only responsible for recommendation selection. -The exact number of paying customers has not been disclosed.

Cost advantage

Cost layer Public information Practical implications
C-side/individual Not mainly object-oriented More focused on team and platform capabilities
API/Developer 10,000 free routing recommendations per month Low routing costs for small-scale projects
Enterprise level Official website FAQ claims more than 30%+ cost savings Actual effects require internal indicator system control to be reproduced on a large scale

The hidden benefit is to upgrade model selection from manual experience to configurable strategies; the hidden cost is the need to build an internal evaluation index system, otherwise it will be difficult to stably optimize the routing strategy.

Main functions

  • Model Routing: Distribute requests to more matching models based on task characteristics.
  • Policy Configuration: Routing policies can be configured according to quality, cost, and delay targets.
  • Evaluation Closed: Optimize subsequent model assignments through result evaluation.
  • Unified Access Layer: Reduce the application layer to be directly bound to a single model supplier.

Model and version evolution

Phase Time Public changes
Prompt Optimization GA 2026-01-20 Prompt Optimization is officially available for general use, routing and prompt word optimization form dual main lines of products
Coding Agent ROI methodology released 2026-06-10 The official blog released the ROI evaluation methodology for coding agents, a non-product version, reflecting the evolution of platform positioning
Routing API made public 2024 Core routing SDK made public, supporting routing recommendations between mainstream models such as OpenAI and Anthropic

Technical advantages

  • Mechanism: The routing layer does not intercept model requests and only recommends model selection; the actual request is still executed by the client through its own gateway, and the data does not pass through a third party.

    Effectiveness: Reduce data privacy risks without changing the existing LLM calling structure. Applicable scenarios: Enterprises with high data sensitivity or teams that focus on data compliance first.

  • Mechanism: Evaluation feedback drives routing optimization. After actual access, the routing strategy will be continuously adjusted based on the execution results.

    Effect: Gradually approach the balance point between cost and quality without relying on manual selection experience. Applicable scenarios: Online applications with large request volume and sensitive indicators.

  • Mechanism: Passed SOC 2, ISO 27001 certification, supports VPC deployment and Zero Data Retention (ZDR) policy.

    Effectiveness: Enterprise customers can introduce a routing layer within the compliance framework without building an additional internal security mechanism. Applicable scenarios: Finance, medical and other industries with high compliance requirements.

How to use

Entrance Applicable roles Recommended steps
Official website Product and technology leader Evaluate capability boundaries and access methods
Docs Platform Engineer Configure routing strategy and API according to documentation
Business Portal Enterprise Procurement Confirm SLA, Compliance and Contract Terms

It is recommended to do A/B on a single scenario task first, and compare the quality, cost, and delay differences between the fixed model and the routing strategy.

Product Pricing

Pay-as-you-go (public volume billing)

Capacity Free quota Exceeding unit price
Intelligent Routing 10,000 times/month $10 / each additional 10,000 times
Prompt Optimization 10 successful optimizations/month $20/each additional successful optimization
Custom Routers 3 free On an extra basis
Target Models supported per run 4 Fixed quota

Enterprise (customized file)

Content Description
Price Contact sales, no public pricing yet
BYO Models Support bringing your own custom models
Priority API Queue Priority Queue Position
Bulk Pricing Scale preferential unit price
Compliance Certification SOC 2, ISO 27001, VPC Deployment ZDR Strategy

Application scenarios

  • Multi-model production application: Automatically select models between different tasks.
  • Cost-sensitive AI products: Use routing to reduce overpriced model abuse.
  • Platformized AI middle platform: Unified management of multiple model supplier access.

Applicable people

  • AI Platform Engineering Team: Need to unify model call governance.
  • Product Team in the Growth Period: Costs need to be controlled after the number of requests increases.
  • Enterprise Technical Leader: Hope to reduce the risk of single supplier binding.

Inappropriate boundary: call small projects with very small scale and long-term model strategy unchanged.

Summary and Outlook

Not Diamond is suitable for mid- and late-stage teams with multi-model parallelism, and can upgrade model selection from "manual experience" to "configurable strategy". The product positioning is clearly routing layer rather than gateway, and the actual requests are still executed within the customer. Current public billing includes 10k free route recommendations per month, and multi-model switching costs are very low. Enterprise-level support for teams with high compliance requirements through SOC 2, ISO 27001, VPC and ZDR policies. The limitation is that the final effect is highly dependent on the perfection of the internal evaluation indicator system.

Related tools: GitHub Copilot, Cursor

Reference sources

Version Info

  • Prompt Optimization officially released (GA) ：The Prompt Optimization function is officially available, and routing and prompt word optimization form the dual main lines of the product.
  • Routing API publicly launched ：The Not Diamond Routing API is public, and developers can use the SDK to recommend multi-model routes.

User Reviews

  • Loading reviews...