The Token Archive

The Token
Archive

Pay less. Ship the same.

Shrink inputs. Compress history. Keep meaning — before tokens hit the model.

Try the pipelineSee pricing

Pricing

Pay less. Ship the same.

Free

$0

API spend under $500/mo

Solo builders · try before you pay

  • Compress + SDK
  • Savings dry-run
  • No card
Sign up free

Starter

$49/mo

$500–$2k API spend/mo

Solo API · OpenAI & Anthropic

  • Drop-in proxy
  • Live metering
  • 3× ROI floor
Sign up to upgrade

Growth

$199/mo

$2k–$8k API spend/mo

Teams · invoice owners

  • Cost dashboard
  • Token reconciliation
  • Team seats
Sign up to upgrade

Enterprise

Custom

$8k+ · or 4% of spend

Scale · compliance · self-host

  • Spend caps
  • SSO / DPA
  • Dedicated support
Talk to us

Four gates.
One thinner prompt.

Each stage cuts waste before assembly — so context stays useful, not bloated.

  1. 01

    Prompt Compression

    Strips filler, hedges, and redundant phrasing without dropping intent.

  2. 02

    Semantic History Cache

    Summarizes older turns so raw chat logs stop eating the window.

  3. 03

    Dynamic Variable Injection

    Pulls only the document chunks that match the query—not the whole dump.

  4. 04

    System Prompt Densify

    Collapses verbose instructions into tight markdown or code-like rules.

Token Archive API

Compress endpoint and OpenAI-compatible chat proxy. Assistant messages stay untouched for prompt-cache safety.

Hit Compress to try Token Archive API

Gate playground

Run compression, history summarization, selective injection, and system densify on a sample — or paste your own.

Run the pipeline to see compressed system rules, summarized history, selective chunk injection, and token savings.