Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Relace / Relace

Relace Compact

Context-compaction model that shortens coding-agent traces while preserving their message format.

Prompt processing
availableClosed weightsPrompt processing
Official documentation (opens in a new tab)Reviewed 15 Oct 2026
4 of 16 specification fields collected
Prompt character limitNot collectedAwaiting source review
Component outputNot collectedAwaiting source review
Input billingSee ratesRelace API
Output billingSee ratesPublished units and conditions below

Specifications

Provider-documented facts. “Not collected” means we have not verified that field yet, not that the feature is unsupported.

Context / input limit
Not collected

No exact model-specific contract established. Unknown is not unsupported.

Maximum input tokens
Not collected

No exact model-specific contract established. Unknown is not unsupported.

Maximum output
Not collected

No exact model-specific contract established. Unknown is not unsupported.

Parameters
Not collected

No exact model-specific contract established. Unknown is not unsupported.

Release date
2026-07-13Source for Release date (opens in a new tab)
Knowledge cutoff
Not collected

No exact model-specific contract established. Unknown is not unsupported.

Reasoning
Not collected

No exact model-specific contract established. Unknown is not unsupported.

Tool calling
Not collected

No exact model-specific contract established. Unknown is not unsupported.

Structured output
Not collected

No exact model-specific contract established. Unknown is not unsupported.

Streaming
Not collected

No exact model-specific contract established. Unknown is not unsupported.

Batch processing
Not collected

No exact model-specific contract established. Unknown is not unsupported.

Fine-tuning
Not collected

No exact model-specific contract established. Unknown is not unsupported.

Context caching
Not collected

No exact model-specific contract established. Unknown is not unsupported.

Input modalities
TextSource for Input modalities (opens in a new tab)
Output modalities
Compressed messagesSource for Output modalities (opens in a new tab)
Generation controls
target_tokens defaults to 96,000 retained agent-model tokens. Counts are heuristic and may differ from the original provider. agent_model identifies the source agent, not a Relace model selector.Source for Generation controls (opens in a new tab)

Access & pricing

Prices belong to an endpoint and billing tier. A provider listing does not guarantee access for every account or region.

Relace API

Provider source (opens in a new tab)
API model ID
Not collected

This task endpoint has no required model selector; agent_model names the model that produced the input trace.

Access
messages required. Chat Completions, Responses and Messages input formats are detected automatically.
Regions
Serving-region inventory not established.
API endpoint
POST https://models.relace.ai/v1/code/compactSource for API endpoint (opens in a new tab)

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 15 Oct 2026
CategoryUsageRateUnit
Historical indexed pricingInput tokens1$0.201M tokens
Historical indexed pricingOutput tokens1$0.201M tokens

Rate conditions

  1. Older indexed pricing page. Current live pricing no longer lists this model, and the authenticated model catalogue was not accessible. Not used as a current quote.
Pricing source (opens in a new tab)

Data & enterprise terms

API prompt and output content is processed in volatile memory, with short-lived KV caching. It is not persisted or used for training/evaluation. Billing and usage metadata is retained; Jacq and other products are outside this policy. Written agreements may differ.

Read the service policy (opens in a new tab)

Caching

Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.

01 / INITIAL INPUTProcess & retain

The system processes eligible input and retains reusable state.

02 / LATER REQUESTCheck available state

Reuse depends on matching input, session state and the cache policy.

03 / CACHE HITReuse & generate

Reusing state can reduce repeated work. Any charges follow the billing rules below.

Relace API

Not verified

No model-specific cache support or discounted cache tariff established.

How it works
Platform policy describes volatile-memory prefix KV caching. Exact model-specific eligibility is unverified.Source for How it works (opens in a new tab)
Minimum reusable input
Not collected

No model-specific cache support or discounted cache tariff established.

Lifetime & refresh
Evicted after inactivity or memory pressure; no numeric TTL guaranteed.Source for Lifetime & refresh (opens in a new tab)
How to enable it
Not collected

No model-specific cache support or discounted cache tariff established.

Usage fields
Not collected

No model-specific cache support or discounted cache tariff established.

Billing rules
Not collected

No model-specific cache support or discounted cache tariff established.

Limits & data boundaries
API content only; billing metadata and other products are excluded from zero-retention scope.Source for Limits & data boundaries (opens in a new tab)

No verified cache rate for this endpoint. An empty rate is not a zero charge.

Caching reviewed 15 Oct 2026

What the measurements say

No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.

Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.

The record behind the model

Why it is included

Provider describes its own trained specialized models, delivered through licensed API services whose IP is retained. The training-method article does not release model weights; the API schema MIT label applies to the schema, not the model.

Classification evidence (opens in a new tab)

Sources reviewed