Adola | Rose 1 prompt compression

Rose 1 production benchmarks are liveView API docs

Semantic compression

Keep what matters.

Rose 1 trims noisy context before your model call and keeps the answer intact.

The customer reported checkout latency after the new billing rollout. Duplicate notes from the support thread repeated the same timeline twice. The incident began when the payments worker retried stale jobs and exhausted the queue. Old search snippets and unrelated ticket history added no evidence. Keep the schema, policy exception, account tier, and citation trail so the model can answer safely.

from adola import Adola

client = Adola(api_key="rose_...")
result = client.compress(
    input=open("retrieved_context.txt").read(),
    query="Which incident caused latency?",
    compression={"target_ratio": 0.3},
    include_spans=False,
)

compressed = result["output"]
receipt = result["receipt"]

Quickstart.

Compress context before the model call. Use the returned text in your next request.

Open docs

Quality first, savings second.

Rose 1 cuts context hard while keeping answers stable across reasoning, science, and math checks.

Rose 1 benchmark pass6 evals

Hard questions, shorter prompts.

30%Typical prompt kept

6Eval sets shown

5Same-accuracy results

2%Largest drop

Latest hard evals

No measured drop with 70% compression.

AIMECompetition math

0% decrease

same accuracy

GPQA DiamondExpert science QA

0% decrease

same accuracy

GDPval-AAProfessional tasks

0% decrease

same accuracy

ARC-ChallengeGrade-school science

2% decrease

near match

CommonsenseQACommonsense reasoning

0% decrease

same accuracy

GSM8KGrade-school math

0% decrease

same accuracy

The production shell is already wired.

Use Adola where context piles up: agent traces, retrieval, prompt gateways, and support copilots. The same workspace gives those flows keys, receipts, billing, and deployable services.

Agent traces

Trim long tool transcripts before the next planning step.

RAG retrieval

Shrink over-retrieved chunks while keeping the answer-bearing spans.

Prompt gateways

Add a compression hop without changing model providers.

Support copilots

Compress ticket history, policy docs, and account context.

For teams that need smaller prompts without turning the model blind.

Cut prompt cost before launching a support copilot.Measured compression ratio by API key across production services.Protected compliance instructions while shrinking retrieved context.Kept model-provider choice open with a simple pre-model API.Cut prompt cost before launching a support copilot.Measured compression ratio by API key across production services.Protected compliance instructions while shrinking retrieved context.Kept model-provider choice open with a simple pre-model API.

Start compressing.

Create a workspace, issue a project key, run the playground, and measure what comes out.

Create workspace View pricing