The Token Company (YC W26) (@thetokenco) / X

The Token Company (YC W26)

31 posts

The Token Company (YC W26)

@thetokenco

Compress long and bloated LLM inputs to save on costs and improve LLM performance

San Francisco, US

Joined January 2026

The Token Company (YC W26)
@thetokenco
Jun 9
Replying to @thetokenco
We have two variants available: bear-2-safety-prompt, for user prompt moderation, and bear-2-safety, for user prompt and agent response moderation. If you’re interested, we're rolling out access to select partners upon request. Read our blog post here:
Introducing Bear-2-Safety
From thetokencompany.com
63
The Token Company (YC W26)
@thetokenco
Jun 9
Replying to @thetokenco
bear-2-safety is a fully extractive compressor, it retains surviving tokens in their original order without any paraphrasing. To implement bear-2-safety, just run one compression model call in front of Llama-Guard, ShieldGemma, or any other safety classifier.
77
The Token Company (YC W26)
@thetokenco
Jun 9
Replying to @thetokenco
Does compression reduce safety classification accuracy? Across five safety benchmarks: OpenAI Moderation, WildGuard, Aegis, BeaverTails, and PKU-QA, F1 retains quality across SoTA classifiers.
70
The Token Company (YC W26)
@thetokenco
Jun 9
Replying to @thetokenco
Our core approach targets asymmetric retention. The vast majority of input tokens don’t influence safety classification pipelines, so we focus our token allocation on problematic spans. Potentially harmful content is maximally retained for safety classifiers, while benign filler
99
The Token Company (YC W26)
@thetokenco
Jun 9
We’re introducing bear-2-safety for private preview, a compression model to optimize cost and performance for LLM safety moderation pipelines, such as Llama Guard and ShieldGemma. Bear-2-safety selectively deletes tokens before safety inference, lowering costs while maintaining
814
The Token Company (YC W26)
@thetokenco
May 23
Our whole research team got banned from Claude Code
721
The Token Company (YC W26)
@thetokenco
May 22
👀👀
Hedgie
@HedgieMarkets
May 21
🦔Microsoft canceled its internal Claude Code licenses this week after token-based billing made the cost untenable, even for a company with effectively infinite cloud resources. Uber's CTO sent an internal memo warning the company burned through its entire 2026 AI budget in just
621
The Token Company (YC W26)
@thetokenco
May 14
Save on your LLM bill with @opencode and @thetokenco
ricky
@drf0k
May 13
pushed an update to the @opencode plugin using the @thetokenco compression model for reducing input tokens for every prompt. i've been meaning to ship an update every since plugins got updated! > added compression status to sidebar widget > added opencode command for easier
00:00
387
The Token Company (YC W26) reposted
The Token Company (YC W26)
@thetokenco
May 13
The Token Company is now HIPAA compliant! Customers using The Token Company's models to compress LLM inputs can now securely process protected health information. Our compression models reduce bloated context from LLM inputs making LLM models perform better and cheaper under
426
The Token Company (YC W26)
@thetokenco
May 13
The Token Company is now HIPAA compliant! Customers using The Token Company's models to compress LLM inputs can now securely process protected health information. Our compression models reduce bloated context from LLM inputs making LLM models perform better and cheaper under
426
The Token Company (YC W26) reposted
Rasmus Uusipaikka
@rasmus_up
May 7
Claude Code making really pretty animations
00:00
203
The Token Company (YC W26)
@thetokenco
Mar 12
X right now
539
The Token Company (YC W26)
@thetokenco
Mar 12
Replying to @thetokenco
Try out the before and after at app.thetokencompany.com
294