Press & Media Kit

Tapas Brand Assets

Everything you need to write about, feature, or reference Tapas — logos, colours, typography, and approved copy.

Press enquiries:

[email protected]

Logo

Tapas

BETA

Dark background (primary use)

Tapas

BETA

Light background

Icon mark (app icon / favicon)

Usage guidelines: Do not recolour the logo, add drop shadows, stretch, or place on busy backgrounds. Maintain a clear space of at least the icon width on all sides. The wordmark uses Inter Bold.

Colour Palette

Brand Green

Primary / CTA

Cyan Accent

Gradient mid / links

Purple Accent

Gradient end / highlights

Background

Dark background

Card Surface

Card / panel surface

Foreground

Primary text

Muted Text

Secondary / caption text

Gradient (CTA):

Typography

Primary typeface

Inter

Used for all UI text, headings, and body copy

Bold (700) — Headlines, CTAs, brand name

Semibold (600) — Section titles, labels

Medium (500) — Navigation, badges

Regular (400) — Body text, descriptions

Monospace typeface

JetBrains Mono

Used for code snippets, API keys, and terminal output

Medium (500) — Inline code

Regular (400) — Code blocks

Approved Boilerplate

Short (1 sentence)

Tapas is an AI energy-efficiency layer that reduces inference costs by up to 99% per query through intelligent semantic caching and question categorisation. It works as a drop-in proxy for any AI platform — OpenAI, Anthropic, Gemini, and more.

Long (3 paragraphs)

Tapas is an AI energy-efficiency platform built on the insight that most AI queries are semantically similar to ones already answered. By routing questions through a semantic cache before reaching a large language model, Tapas reduces energy consumption by up to 99% per query while returning answers in milliseconds. The platform supports 160+ knowledge categories across science, medicine, finance, programming, and more. It integrates with any AI provider via a drop-in base URL swap and ships SDKs for TypeScript, Python, Go, and Rust. Tapas is available as a standalone API and as middleware for existing AI pipelines. Founded on the principle that intelligence should be efficient — like the Spanish tapas tradition of small, satisfying bites — Tapas delivers precision without waste.

Key Facts

Up to 99%

Energy reduction per query vs. full LLM inference

160+

Knowledge categories across 7 domains

5 SDKs

TypeScript, Python, Go, Rust, OpenAI-compatible

< 5 ms

Median cache-hit response latency

Drop-in

Base URL swap — no code changes required

Open API

REST + tRPC, documented at tapas.one/docs/rest

Press enquiries

For interviews, fact-checking, embargoed announcements, or high-resolution assets, contact our press team directly.