# The Token Company

Source: [The Token Company](<https://www.sota2.com/companies/the-token-company>)

Collection index: [llms.txt](<https://www.sota2.com/companies/llms.txt>)

Data updated: 2026-07-24T20:37:55.592Z

Provider of a prompt compression API that strips low-signal tokens from LLM inputs to reduce API costs across GPT, Claude, and Gemini.

Compression middleware that removes context bloat in milliseconds, lowering costs and improving end-to-end latency. Compression is especially effective across natural language workloads. In a blind LLM arena case study with one of our customers, compressed requests increased user preference, lowered costs, and lifted purchase volume by 5%.

## Company facts

| Field | Recorded value |
| --- | --- |
| Local name | Not recorded |
| Categories | Models & AI Infrastructure |
| Headquarters | San Francisco, US |
| Deployment | Not recorded |

## Products by this company

- [Bear-2 Prompt Compression API](<https://www.sota2.com/products/the-token-company-bear-2-prompt-compression-api.md>): Prompt compression API featuring the bear-2 model that strips low-signal tokens from raw LLM inputs such as documents, websites, and transcripts before they hit the LLM context...

## Task-specific ranking appearances

## Customer stories and references

These stories and references are supplied or named by the company.

- [Customer case](<https://thetokencompany.com/blog/pax-historia>): The Token Company reports integrating bear-1.1 into Pax Historia's conversation pipeline to reduce context costs. The vendor reports higher user preference for compressed variants in its model arena and a 5% increase in…

- [Customer case](<https://thetokencompany.com/blog/helonic>): The Token Company reports that Helonic uses bear-1.2 between extraction and reasoning to compress construction-drawing prompts. At a conservative 0.05 aggressiveness setting, the vendor reports approximately 47,000…

## Original provider information

[Official website](<https://thetokencompany.com/>)
