- Instruction Following
Qwen3.5 397B A17B is a large-scale language model from Qwen with roughly 397 billion parameters, designed for advanced reasoning and multilingual understanding. It targets high-end inference…
Powered by Upstage
Solar Pro 3
- Instruction Following
Solar Pro 3 is Upstage’s Mixture-of-Experts large language model with 102B total parameters (12B active), a 128K-token context window, and strong extended reasoning and tool-use capabilities.
Providers Comparison
Speed
Output tokens per second · Higher is better
L
LLM API
S
SiliconFlow (FP8)
G
GMI
C
CoreWeave
N
Novita
D
DeepSeek
P
Parasail (FP8)
D
DeepInfra (FP4)
End-to-End Response Time
Seconds · Lower is better
L
LLM API
S
SiliconFlow (FP8)
G
GMI
C
CoreWeave
N
Novita
D
DeepSeek
P
Parasail (FP8)
D
DeepInfra (FP4)
Price
USD per 1M tokens (blended) · Lower is better
L
LLM API
S
SiliconFlow (FP8)
G
GMI
C
CoreWeave
N
Novita
D
DeepSeek
P
Parasail (FP8)
D
DeepInfra (FP4)
About the model
What is Solar Pro 3?
Solar Pro 3 is a proprietary Mixture-of-Experts language model from Upstage optimized for efficient, high-quality text generation and reasoning. It is used for complex multi-step reasoning, agentic workflows, and long-context tasks such as document analysis and large-codebase assistance. It also serves enterprise applications that need reliable tool use, structured outputs, and multilingual support focused on Korean with additional English and Japanese coverage. Solar Pro 3 follows earlier Solar-series models such as Solar Pro 2, offering increased parameter scale and improved reasoning performance within the same general model family.
Model capabilities
5 Core Capabilities
-
Text Generation
Generates and edits high-quality text responses across domains, suitable for content creation, SEO workflows, and structured writing tasks.
-
Long Context Handling
Processes and reasons over long inputs with a context window up to 128K tokens, supporting document-heavy and retrieval-oriented applications.
-
Tool Use
Supports tool use and function calling, enabling agentic workflows that interact with external systems and APIs programmatically.
-
Structured Outputs
Produces well-structured JSON and schema-conformant outputs, useful for automation pipelines and programmatic integration with downstream systems.
-
Multilingual Support
Handles multiple languages with strong performance in Korean and solid English and Japanese support for multilingual applications.
Use cases
6 Most Valuable Use Cases
- Code Generation Assistance
- Enterprise Document Search
- Contract Review Support
- Invoice Extraction Automation
- Customer Support Triage
- Business Process Workflows
Transparent pricing
Cost Comparison
LLM API offers the lowest cost and latency for Solar Pro 3–class models.
| Provider | Region | Latency | Throughput | Uptime | Input ($/1M) | Output ($/1M) | Context |
|---|---|---|---|---|---|---|---|
| LLM API BEST | Global | 120ms | 80 tps | 99.99% | $0.20 | $0.60 | 256K |
| Upstage | Global | ~250ms | ~40 tps | ~99.9% | ~$0.25 | ~$0.75 | ~128K |
| OpenRouter | Global | ~320ms | ~35 tps | ~99.9% | ~$0.30 | ~$0.90 | ~128K |
| Together AI | US East | ~280ms | ~45 tps | ~99.9% | ~$0.28 | ~$0.85 | ~128K |
| Fireworks AI | US West | ~260ms | ~50 tps | ~99.95% | ~$0.26 | ~$0.80 | ~128K |
Performance benchmarks
Technical Specifications
| Metric | Solar Pro 3 (Upstage) | GPT-4.1 (OpenAI) | Claude 3.5 Sonnet (Anthropic) |
|---|---|---|---|
| Avg Latency | ~180ms | ~220ms | ~250ms |
| Context Window | 200K | 128K | 200K |
| Input Price ($/1M) | $0.80 | $5.00 | $3.00 |
| Output Price ($/1M) | $4.00 | $15.00 | $15.00 |
| Max Output Tokens | 4K | 4K | 4K |
| Throughput | ~80 tps | ~60 tps | ~50 tps |
| Uptime | 99.9% | 99.9% | 99.9% |
30-day usage via LLM API
- 9.8B
- Prompt tokens processed (last 30 days)
- 7.1B
- Completion tokens generated (last 30 days)
- 12.4M
- API requests served (last 30 days)
- 99.8%
- Average API uptime
Architecture & Integration
Why Build on LLM.API?
One unified API. Every major model. Built-in reliability, cost control, and observability.
-
Unified AI Routing
Dynamically route each request to the optimal model across providers using latency, cost, and quality signals—without changing your integration or redeploying code.
One endpoint, every model. -
Smart Cost Controls
Define budgets, price ceilings, and routing rules so LLM.API automatically picks the cheapest viable model while preserving output quality and performance SLAs.
Optimize spend by default. -
Automatic Fallbacks
Guard against provider outages and rate limits with configurable failover logic that instantly retries on backup models, maintaining uptime without custom error-handling glue.
Resiliency built in. -
Deep Observability
Get per-request traces, latency and cost breakdowns, and structured logs across all providers from a single dashboard and API, ready for alerting and analytics.
One pane of glass. -
Task-Aware Orchestration
Describe tasks at a higher level—chat, extraction, tools—and let LLM.API select prompts, parameters, and models, so you ship features instead of tuning configs.
Think tasks, not prompts. -
High-Throughput Batch
Send massive batches of jobs through a single API call with concurrency controls, retries, and progress tracking, ideal for backfills, evaluations, and bulk processing.
Scale jobs, not code.
Decision guide
When to Use — When NOT to Use
Use it if...
- You need a strong general-purpose model for chatbots and virtual assistants.
- You need solid coding assistance, including code completion, debugging, and explanation tasks.
- Your use case involves multilingual text understanding and generation across many major languages.
- Your use case involves drafting, rewriting, and polishing emails, reports, and marketing copy.
- You need a capable model for question answering over moderately long documents or webpages.
- You need a balance between quality and cost for everyday enterprise productivity workflows.
Avoid if...
- You need cutting-edge performance on the hardest reasoning or math competition benchmarks.
- Your workload requires guaranteed support for extremely long contexts, like hundreds of thousands tokens.
- You need tightly integrated image or multimodal capabilities beyond basic text-only interactions.
- You need deterministic, fully reproducible outputs with strict token-by-token compatibility guarantees.
- Your workload requires highly specialized domain models, like medical diagnosis or legal argumentation.
- You need robust offline deployment on highly constrained edge devices with minimal hardware resources.
FAQ
Frequently Asked Questions
-
What is Solar Pro 3?
Solar Pro 3 is a large language model by Upstage optimized for high-quality reasoning, coding, and general-purpose chat via the LLM.API gateway.
-
What is Solar Pro 3 best used for?
Solar Pro 3 is best for complex reasoning, code generation and debugging, multi-step tool use, and production-grade chatbots needing strong instruction following.
-
What is the context window of Solar Pro 3?
Solar Pro 3 supports a long context window suitable for large documents and multi-step conversations; check the LLM.API model card for the exact token limit.
-
How fast is Solar Pro 3 in terms of latency and throughput?
Typical end-to-end latency is on the order of seconds for short prompts, with streaming responses and scalable throughput handled by LLM.API infrastructure.
-
Which modalities does Solar Pro 3 support?
Solar Pro 3 is a text-only model that accepts text prompts and returns text completions or chat responses.
-
How do I call Solar Pro 3 through LLM.API?
You can select the upstage/solar-pro-3 model name in the LLM.API completion or chat endpoint, passing your prompt and any temperature or max_tokens parameters.
-
How is Solar Pro 3 priced on LLM.API?
Solar Pro 3 uses pay-as-you-go, per-token billing; see the LLM.API pricing page for current input and output token rates.
-
How does Solar Pro 3 compare to similar models?
Solar Pro 3 is positioned as a high-quality, cost-efficient general model competitive with other top-tier reasoning and coding LLMs in its price bracket.
-
What are the main limitations of Solar Pro 3?
Solar Pro 3 can hallucinate, lacks real-time knowledge or browsing, and may underperform on highly specialized domain tasks without careful prompting or grounding.
-
Can I fine-tune or customize Solar Pro 3 via LLM.API?
Fine-tuning support depends on LLM.API capabilities at the time; check the model page for whether custom fine-tunes or adapters are available for Solar Pro 3.
Compare
Competitive Models
- Instruction Following
Free Models Router is an OpenRouter meta-model that automatically routes requests to compatible free models, providing no-cost inference across multiple underlying LLMs. It filters candidates based…
- Instruction Following
GPT-5.3 Chat is an OpenAI conversational large language model designed for general-purpose dialogue and task assistance, with improved reasoning and instruction-following over prior GPT chat models.