Solar Open 100B

SOLAR

Open WeightModel weights are publicly available — can be downloaded and self-hosted

Released December 2025131K context≈ 98,304 words100B params

Solar Open 100B carries itself like a specialist who punches above its weight class — a 100B parameter open-weight model built around Upstage's depth-upscaling technique, which stitches together layers from smaller models rather than training from scratch. It handles long documents comfortably with a 131K token context window. The trade-off is that its unconventional architecture can behave differently from standard transformer stacks, so fine-tuning or deployment may require extra attention.

Capabilities

Capability scores are AI-generated based on model documentation, benchmarks, and technical specifications. Learn more

Factual Knowledge

Moderate

Reasoning & Logic

Solar Open 100B

SOLAR

Open WeightModel weights are publicly available — can be downloaded and self-hosted

Released December 2025131K context≈ 98,304 words100B params

Solar Open 100B carries itself like a specialist who punches above its weight class — a 100B parameter open-weight model built around Upstage's depth-upscaling technique, which stitches together layers from smaller models rather than training from scratch. It handles long documents comfortably with a 131K token context window. The trade-off is that its unconventional architecture can behave differently from standard transformer stacks, so fine-tuning or deployment may require extra attention.

Capabilities

Capability scores are AI-generated based on model documentation, benchmarks, and technical specifications. Learn more

Factual Knowledge

Moderate

Reasoning & Logic

Glossary

ArchitectureThe underlying structural design of a neural network that defines how data flows through layers and components.AttentionA mechanism that lets the model focus on relevant parts of the input when generating each output token.Context WindowThe maximum number of tokens a model can process in a single conversation or prompt.Depth-UpscalingA technique that creates a larger model by combining and stitching together layers from smaller pre-trained models rather than training a new model from scratch.Fine-TuningThe process of further training a pre-trained model on new data to adapt it for specific tasks or domains.Open-Weight ModelA model whose trained weights are publicly released, allowing anyone to download and run it locally.TokenA small unit of text (a word, subword, or punctuation mark) that a language model breaks input into for processing.TransformerThe dominant neural network architecture for language models, using self-attention to process sequences.UpscalingUsing sparse local measurements to estimate values across a larger geographic or temporal region.

Capabilities

Capabilities

Use Case Fit

Glossary