Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Upstage / Solar Embedding

Solar Embedding 2 Query

Purpose-specific query embeddings for semantic retrieval.

Embeddings
availableClosed weightsEmbeddings
Official documentation (opens in a new tab)Reviewed 14 Oct 2026
11 of 23 specification fields collected
Input token limit8K8,000 tokens
Vector dimensions1K1,024 dimensions
Input billingSee ratesUpstage 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
8,000 tokensSource for Context / input limit (opens in a new tab)
Maximum input tokens
8,000 tokensSource for Maximum input tokens (opens in a new tab)

Per-text context limit; batch request limits are separate.

Maximum output
Not collected

Not verified for this exact release or deployment. No unrelated sibling specification inherited.

Parameters
Not collected

Not verified for this exact release or deployment. No unrelated sibling specification inherited.

Architecture
Not collected

Not verified for this exact release or deployment. No unrelated sibling specification inherited.

Languages
Not collected

Not verified for this exact release or deployment. No unrelated sibling specification inherited.

Knowledge cutoff
Not collected

Not verified for this exact release or deployment. No unrelated sibling specification inherited.

Release date
2026-06-30Source for Release date (opens in a new tab)
Reasoning
Not collected

Not verified for this exact release or deployment. No unrelated sibling specification inherited.

Tool calling
Not collected

Not verified for this exact release or deployment. No unrelated sibling specification inherited.

Structured output
Not collected

Not verified for this exact release or deployment. No unrelated sibling specification inherited.

Streaming
Not collected

Not verified for this exact release or deployment. No unrelated sibling specification inherited.

Batch processing
Not collected

Not verified for this exact release or deployment. No unrelated sibling specification inherited.

Fine-tuning
Not collected

Not verified for this exact release or deployment. No unrelated sibling specification inherited.

Context caching
Not collected

Not verified for this exact release or deployment. No unrelated sibling specification inherited.

Input modalities
TextSource for Input modalities (opens in a new tab)
Output modalities
EmbeddingsSource for Output modalities (opens in a new tab)
Release identity
solar-embedding-2-querySource for Release identity (opens in a new tab)

Aliases are endpoint selectors attached to the identified version, not additional independent model records.

Embedding dimensions
1,024 dimensionsSource for Embedding dimensions (opens in a new tab)
Request limits
Up to 100 texts and 204,800 total tokens per requestSource for Request limits (opens in a new tab)

Synchronous arrays of inputs do not establish an asynchronous discounted Batch API.

Vector normalization
Unit-length normalized vectors; dot product equals cosine similaritySource for Vector normalization (opens in a new tab)
Embedding value types
Floating-point vectorsSource for Embedding value types (opens in a new tab)
Base model
Fine-tuned from microsoft/harrier-oss-v1-0.6bSource for Base model (opens in a new tab)

Public base-model weights do not establish public release of the provider fine-tune.

Access & pricing

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

Upstage API

Provider source (opens in a new tab)
API model ID
solar-embedding-2-query

Exact provider-listed version selector; current availability depends on lifecycle.

Access
API key/account required. Private and marketplace deployments have separate access and billing.
Regions
Not collected

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 14 Oct 2026
CategoryUsageRateUnit
Embedding inputInput tokens1$0.021M tokens

Rate conditions

  1. Direct API embedding input rate. No generated-output price or text-question estimate is inferred; VAT excluded.
Pricing source (opens in a new tab)

Data & enterprise terms

Retention, training use, regional processing and enterprise terms have not been summarized for this endpoint.

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.

Upstage API

solar-embedding-2-query
Not verified

No verified exact-model cache contract for this field. Missing documentation does not establish lack of support.

How it works
Not collected

No verified exact-model cache contract for this field. Missing documentation does not establish lack of support.

Minimum reusable input
Not collected

No verified exact-model cache contract for this field. Missing documentation does not establish lack of support.

Lifetime & refresh
Not collected

No verified exact-model cache contract for this field. Missing documentation does not establish lack of support.

How to enable it
Not collected

No verified exact-model cache contract for this field. Missing documentation does not establish lack of support.

Usage fields
Not collected

No verified exact-model cache contract for this field. Missing documentation does not establish lack of support.

Billing rules
Not collected

No verified exact-model cache contract for this field. Missing documentation does not establish lack of support.

Limits & data boundaries
Not collected

No verified exact-model cache contract for this field. Missing documentation does not establish lack of support.

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

Caching reviewed 14 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

Closed-weight distribution inference: the exact commercial release is listed by the provider, while the reviewed official public inventory contains no matching trained weights. Public preview/base weights and tokenizer-only repositories do not establish release of this exact commercial checkpoint.

Classification evidence (opens in a new tab)

Lifecycle dates

Provider-listed version release30 Jun 2026Source (opens in a new tab)

Sources reviewed