SOLAR 10.7B Japanese

Released
2024-06-24
Last refreshed
2026-10-13
Status
Researched 151d ago
Open sourceCommercial use: permitted
Upstage releases · 3 in the last 12 monthsChangelog →
Specifications
Released
2024-06-24
Context
4K
Parameters
10.7B
Architecture
Decoder Only
Specialization
general
Openness
Open source
License
Apache 2.0OSI-approvedPermitted
Weights
Unknown
Code
Unknown
Training
Fine-tuned
Created by

Developing AI for interactive entertainment.

Yongin, Ch'ungch'ong-namdo, South Korea
Founded 2020
Website
Pricing
Output / 1M
$0.300
Input / 1M
$0.300

Cheapest of 1 route · Upstage Console

About

The SOLAR-10.7B Japanese language model boasts 10.7 billion parameters and is constructed on the Llama 2 architecture, integrating weights from Mistral 7B. It uses a technique called Depth Up-Scaling (DUS) to scale its depth in a straightforward manner compatible with standard frameworks. Tailored for single-turn conversations, the model excels at generating text and potentially code, rivaling even those with up to 30 billion parameters in certain benchmarks. Although its performance varies across tasks, and the details of its training data are limited, it remains a proficient tool in the realm of language models.

Providers(1)

Compare API pricing across 1 tracked provider route for input and output tokens, batch, and cached reads when available.

ProviderInput / 1MOutput / 1MRoute
Upstage Console$0.300$0.300
Serverless

Capabilities

No model capability flags are currently sourced.

Benchmark peer barsfor Coding

No task-mapped benchmark peers are available for this model yet.

Migration checks

No linked migration route is available for this model yet.