Not Diamond
Not Diamond is an AI infrastructure company that provides an intelligent model router, using a meta-model to automatically route each query to the most cost-effective LLM while optimizing accuracy and latency, serving engineering teams deploying coding agents and multi-model AI applications.
- Company typePrivate
- Founded1980
- HeadquartersSan Francisco, United States
- Headcount1–10
- GTM typeB2B
- OfferingSoftware
What Not Diamond does
Not Diamond is an AI infrastructure company that builds an intelligent model router, a meta-model that predicts which large language model (LLM) is best suited for each incoming query and routes requests accordingly, combining multiple LLMs to improve output quality while reducing inference cost and latency. The platform exposes a REST API with Python and TypeScript SDKs, offers pre-trained routers for chat and coding-agent workloads, supports custom routers trained on customer evaluation data (minimum 15 samples), and includes an automated prompt optimization feature using evolutionary algorithms and reinforcement learning. The core technology stack includes reinforcement learning from feedback signals, Pareto optimization for cost-quality tradeoffs, random forest classifiers for pairwise routing, and privacy-preserving routing that operates on derived metadata without inspecting payload content.
The company's go-to-market combines product-led growth (free tier with 10K routing recommendations per month, 5-minute self-serve integration) with enterprise sales (SAML SSO, MDM, SLAs, 24/7 support) and AWS Marketplace distribution. Pricing is primarily usage-based ($10 per additional 10K recommendations, $20 per additional prompt optimization) with custom enterprise contracts. Not Diamond has accumulated 100,000+ users and 1.5M+ human preference arena samples, and reports customer outcomes including 30%+ cost savings, 5%+ accuracy gains, and a 39% accuracy improvement on Rootly's SRE benchmarks.
Not Diamond serves engineering teams building AI-powered applications, particularly those deploying coding agents and multi-model architectures. Named customers and partners include Hugging Face, Dropbox, IBM, DoorDash, American Express, SAP, Rootly, Replicated, and OpenRouter. Founded in 2023 in San Francisco with a ~15-person team led by CEO Tomás Hernando Kofman, CTO Tze-Yang Tung, and COO Jeffrey Akiki, the company is backed by Defy.vc, Inovia Capital, IBM Ventures, SAP.iO, and individual investors including Jeff Dean (Google), Ion Stoica (Databricks), and Guillermo Rauch (Vercel).
Not Diamond firmographics
Firmographics- Name
- Not Diamond
- Legal name
- Vector
- Website
- https://notdiamond.ai
- Company type
- Private
- Founded year
- 1980
- Operating status
- Operating
- Headcount range
- 1–10 employees
- Short description
- Not Diamond is an AI infrastructure company that provides an intelligent model router, using a meta-model to automatically route each query to the most cost-effective LLM while optimizing accuracy and latency, serving engineering teams deploying coding agents and multi-model AI applications.
- Ownership category
- akta.pro rank
Not Diamond industry classification
Industry- Product category
- AI Infrastructure / LLM Model Routing
- akta.pro primary industry
- Smart Contract Execution Environments (EVM/WASM, runtimes, precompiles) (FSAPABAA)
Keywords
Where Not Diamond is headquartered
LocationHeadquarters
- HQ city
- San Francisco
- HQ country
- United States
- HQ region
- North America
Offices1 record
Markets served
Not Diamond business model
Business model- GTM type
- B2B
- Offering type
- Software
- Cost components
- Technology or R&D, Personnel, Infrastructure, Marketing or Sales, Operations
Revenue model
- Usage-based Routing: Charges per routing recommendation: $10 per 10K additional recommendations after free tier. Revenue based on API usage volume with per-request pricing.
- Prompt Optimization: Charges $20 for each additional successful optimization after free tier (10 free per month). Revenue from prompt optimization service usage.
- Enterprise Licensing: Custom enterprise pricing with organization-wide analytics, advanced security controls, MDM deployments, SSO with SAML, SLAs and 24/7 support. Volume-based pricing for larger deployments.
- AWS Marketplace: Available through AWS Marketplace with private offers, consolidated billing, and pre-approved budgets for enterprise customers.
Pricing tiers
| Model | Billing | Price |
|---|---|---|
| Usage-based | Pay-as-you-go | Pay-as-you-go for developers and small teams |
| Freemium | Monthly | Early access for coding agents |
| Subscription | Annual | Enterprise teams ready to scale |
Go-to-market motion2 records
Distribution channels4 records
Marketing channels7 records
Not Diamond product offering
Product offeringCore offering
Not Diamond provides an intelligent AI model routing platform that combines multiple large language models (LLMs) into a meta-model which predicts, for each incoming query, which underlying model will produce the best response based on quality, cost, or latency tradeoffs. The platform is delivered as a REST API with Python and TypeScript SDKs and includes automated prompt optimization using evolutionary algorithms and reinforcement learning. Pre-trained routers (Chat, Code) are available out of the box, while customers can also train custom routers on their own evaluation data.
Product overview
Not Diamond is an AI infrastructure company offering an intelligent model routing platform with two core product capabilities: (1) the Not Diamond Intelligent Model Router — which predicts the best LLM for each query using a meta-model approach combining multiple LLMs — available as pre-trained routers (Chat for general use, Code for coding agents) and custom routers trained on user evaluation data; and (2) Prompt Optimization — an automated prompt engineering feature using evolutionary algorithms and reinforcement learning to optimize prompts for target models. The platform is accessible via REST API, Python SDK, and TypeScript SDK, with deployments on AWS Marketplace. The company also maintains open-source components including the RoRF pairwise router and notdiamond-0001 model on Hugging Face, and offers a consumer chatbot (chat.notdiamond.ai) with arena mode for human preference collection.
Differentiator
Problem solved
Functional benefit
Products and services
- Not Diamond Intelligent Model Router An intelligent model router that predicts which LLM is best-suited for each incoming query, combining multiple LLMs into a meta-model that learns when to call each one. Improves output quality while reducing costs and latency through dynamic model selection based on quality, cost, or latency tradeoffs. Delivered as a REST API for developers and engineering teams.
- Not Diamond Prompt Optimization An automated prompt optimization feature that uses evolutionary algorithms and reinforcement learning to generate optimized prompts for target models. Improves accuracy and seamlessly adapts prompts to new models without manual engineering, collapsing what typically takes weeks of effort into minutes of automated processing. Available via API for developers and AI teams.
- Pre-trained Router: Chat
Quantifiable outcome
- 5%+ accuracy gains across coding agent workloads
- +6 more outcomes
Companies that use Not Diamond
Customer profileNamed customers9 records
Segments3 records
Ideal customer profiles3 records
Not Diamond technology and API
TechnologyTechnology focussed Yes
API detail
- Has API
- Yes
- API docs
- API detail
Core technology
AI maturity
App detail
Integration2 records
AI capability7 records
Feature7 records
Not Diamond partnerships and signals
Strategic signalPartnerships
Two partnerships are on record, tiered core.
- SAPcoreSAP integrated Not Diamond's prompt optimization into their generative AI hub at Sapphire 2025. Philipp Herzig, CTO and Chief AI Officer of SAP SE, showcased the preview. Partnership enables SAP to drive accuracy improvements on enterprise use cases and accelerate engineering throughput.
- Amazon Web Services (AWS)coreNot Diamond available on AWS Marketplace with Private Offers, enabling AWS customers to seamlessly integrate AI model routing with Not Diamond in existing AWS infrastructure. Provides consolidated billing, pre-approved budgets, and simplified procurement for AWS customers.
Scale indicators9 records
Recent moves6 records
Expansion highlights6 records
Not Diamond competitors and assessment
Company assessmentDirect peers
- Martian: LLM routing and model-merging platform that dynamically routes between models to optimize cost and accuracy. Direct competitor targeting the same developer and enterprise segments as Not Diamond.
- OpenRouter: Unified API that routes across leading LLMs with intelligent model selection, directly competing with Not Diamond's core routing offering. Notably also a Not Diamond customer, illustrating the unusual customer-competitor overlap.
- Portkey: AI gateway platform offering LLM routing, fallbacks, observability, and cost controls to developers and enterprises. Directly comparable to Not Diamond on routing, observability, and enterprise feature set.
- Helicone: Open-source LLM observability platform with built-in routing, caching, and cost controls. Targets similar developer-first audience and overlaps directly with Not Diamond's PLG funnel.
- LiteLLM: Open-source LLM gateway and proxy offering routing, fallbacks, and load balancing across 100+ models. Competes on routing and orchestration with a self-hosted-first approach.
Broad incumbents
- Together AI: AI cloud platform providing model serving, fine-tuning, and inference APIs across leading open-source and proprietary models. Competes at the model-access layer that Not Diamond routes between.
- Anyscale: Ray-based AI compute platform from the creators of Ray, offering model serving and inference at scale. Overlaps with Not Diamond as an alternative AI infrastructure layer and is also an investor via Ion Stoica.
- AWS Bedrock: Amazon's managed service providing access to multiple foundation models with built-in routing and orchestration. Represents the incumbent cloud-provider alternative Not Diamond must compete against in enterprise deals.
- LangChain: AI orchestration framework for building LLM applications with model selection, routing, and chaining capabilities. Adjacent competitor in the AI developer tooling layer with broader scope than pure routing.
Emerging players
- Unify: LLM routing and benchmarking platform that lets developers compare and route across multiple models with cost and latency optimization. Comparable on routing functionality though smaller scale.
Market position
Strengths5 records
Weaknesses5 records
Competitive moat5 records
Key risks6 records
Key highlights7 records
Customer concentration
Not Diamond compliance and trust
Trust signalCompliance4 records
Not Diamond financial estimates
Financial estimateRevenue estimate
Valuation estimate
Not Diamond leadership team
Management profileNumber of profiles
Profiles4 records
Not Diamond funding detail
Funding detailFunding overview
Funding rounds2 records
Investors11 records
Funding detail is available on the Subscription and Enterprise plan.Contact sales →
Not Diamond M&A and investment
M&A and investmentM&A
Investments
M&A and investment is available on the Subscription and Enterprise plan.Contact sales →
Frequently asked questions about Not Diamond
What does Not Diamond do?
Not Diamond provides an intelligent AI model routing platform that combines multiple large language models (LLMs) into a meta-model which predicts, for each incoming query, which underlying model will produce the best response based on quality, cost, or latency tradeoffs. The platform is delivered as a REST API with Python and TypeScript SDKs and includes automated prompt optimization using evolutionary algorithms and reinforcement learning. Pre-trained routers (Chat, Code) are available out of the box, while customers can also train custom routers on their own evaluation data.
Is Not Diamond a public or private company?
Not Diamond is a private company. It is classified as venture growth investor backed and is currently operating.
When was Not Diamond founded?
Not Diamond was founded in 1980. It employs 1 to 10 people.
Where is Not Diamond based?
Not Diamond is headquartered in San Francisco, United States, in the North America region.
How does Not Diamond make money?
Four revenue lines are on record. Usage-based Routing is the primary driver. The others are prompt Optimization, enterprise Licensing and AWS Marketplace.
Who are Not Diamond's main competitors?
Direct peers on record are Martian, OpenRouter, Portkey, Helicone and LiteLLM. Broad incumbents are Together AI, Anyscale, AWS Bedrock and LangChain. Unify is listed as an emerging player.
Does Not Diamond have an API?
Yes. Not Diamond offers a public REST API that enables developers to integrate intelligent model routing and prompt optimization into their applications. The API supports: (1) Model Router — predicts which LLM to use for each prompt, with tradeoff options for quality (default), cost, or latency, plus a cost_quality_tradeoff parameter (0-10 scale) for fine-grained blending; (2) Prompt Optimization — automatically generates optimized prompts for target models using evolutionary algorithms and reinforcement learning, with evaluation metrics including LLM-as-a-judge (semantic similarity, custom), RAG-specific (faithfulness, relevance), and computed metrics (JSON Match, Exact Match, BLEU, ROUGE, METEOR). The API returns session IDs for tracking and preference IDs for custom routers. Authentication is via Not Diamond API key (generated at app.notdiamond.ai/keys). Rate limits are 50 requests per second; users can request higher limits by contacting the team. Pre-trained routers and custom routers (trained on user evaluation data with minimum 15 samples, up to 5MB or 10,000 samples per job) are both supported. Custom models can be included in routing decisions by defining context_length, input_price, output_price, and latency parameters. Developer documentation is at docs.notdiamond.ai.
What industry is Not Diamond in?
Not Diamond's product category is AI Infrastructure / LLM Model Routing. Its primary akta.pro industry code is FSAPABAA, Smart Contract Execution Environments (EVM/WASM, runtimes, precompiles).