Your agents pay for tokens they never needed.

Every turn, a coding agent re-sends the whole conversation - files it already read, tool output it already used, its own reasoning. You pay for all of it. SOMA compresses the context and the chain-of-thought on the way to the model, and keeps what the task needs.

Context grows every turn. So does the bill.

  • AAgents re-send history on every request - the same files, the same tool output.
  • BCost per task grows faster than the task itself: turn 20 carries turns 1-19.
  • CMost of that context is no longer needed to get the next step right.

At turn 14

This session has sent 658k tokens so far. 39k of the last request was context the model no longer needed - 45% of it.

Tokens sent per agent turn

NeededRemovable
0k25k50k75k100k13579111314

Illustration - a typical 14-turn coding session. Real figures land with the benchmark release.

Benchmarks

Same agent. Same model. Fewer tokens.

We run the same agent on the same task twice - with compression off and on - and compare cost and whether the task still passed. The router reports both figures: what it measured on your traffic, and our current estimate where measurement is thin.

Agent · ModelSaved
GitHub Copilot CLIDeepSeek V4 Pro
−15%
GitHub Copilot CLIGLM 5.2
measuring
GitHub Copilot CLIDeepSeek V4.1 Flash
measuring
Codexcoming soon
measuring
Claude Codeplanned
measuring
Zedmanual setup
measuring

Estimated savings · today

15%

GitHub Copilot CLI + DeepSeek V4 Pro, our current estimate for this pair. The router labels every figure it shows you as measured or estimated - and never presents one as the other.

What we test on

  1. 1Real issues from real GitHub repositories
  2. 2Navigating large codebases and changing many files
  3. 3Debugging, running and fixing tests
  4. 4Long tasks with many tool calls and reasoning rounds

Savings depend on the task, the context size and how much of it the provider already caches. Long sessions with heavy tool output save the most; short ones can save nothing. Measured per-key figures are in your dashboard.

Agents & models

Keep your agent. Keep your model.

SOMA speaks the OpenAI API, so the agent doesn't have to change. Point it at our endpoint and it starts sending less.

Your agent

GitHub Copilot CLIone-line installer
Live
Any OpenAI-compatible clientbase URL + API key
Live
Codexby OpenAI
Soon
Zedmanual provider setup
Soon
Claude Codeby Anthropic
Planned
Cursorcustom model provider
Planned
OpenClawour open-source compressor
Planned
Compress

The model

DeepSeek V4 ProDeepSeek
Live
GLM 5.2Z.ai
Live
DeepSeek V4.1 FlashDeepSeek
Live
GPT familyOpenAI
Planned
Claude SonnetAnthropic
Planned
Qwen 3 Coderopen weights
Planned
Kimi K2open weights
Planned

Setup

One line, then keep working.

Your key starts with sk-soma. Sign up, copy it, run one command - the agent you already use keeps working, just with less going out.

GitHub Copilot CLIshell
# install Copilot CLI and point it at SOMA
curl -s https://app.thesoma.ai/copilot | bash -s sk-soma-YOUR-KEY
# that's it - start coding
copilot
New accounts $5 in creditsMarkup nonePay with card or TAO

Compression runs before the request leaves us. If the compressor is unavailable, your request goes through uncompressed rather than failing.

Research & development · Bittensor subnet 114

Our compressors come from an open competition.

SOMA runs subnet 114 on Bittensor. Miners compete to build the best compressor, validators score every submission on real work, and after each competition our compression team takes the winning code and adapts it to the app.

  1. 01

    Miners submit

    Independent teams train and submit compressors to the subnet.

  2. 02

    Validators score

    Every submission runs on real tasks. Cost and task success, both.

  3. 03

    Our team adapts it

    After each competition, our compression team takes the winning code and makes it production-ready.

  4. 04

    It ships to the app

    The adapted compressor runs in front of your agent, and usage funds the next round.

Subnet
SN 114
Compresses
Context + CoT
Active miners
-
Leaderboard
public

Get started

Pay for the context that matters.