Solar Pro 3
Upstage · released Jan 27, 2026
Solar Pro 3 is Upstage's powerful Mixture-of-Experts (MoE) language model. With 102B total parameters and 12B active parameters per forward pass, it delivers exceptional performance while maintaining computational efficiency. Optimized...
Specification
- Context window
- 131K
- Max output
- 118K
- Knowledge cutoff
- Not stated
- Parameters
- Undisclosed
- Licence
- Proprietary
- Serving providers
- 1
- Moderated
- No
- Uptime
- 100.0%
Intelligence
7.8
30th percentile
Coding
16.2
Coding Index
Agentic
1.4
Agentic Index
Output speed
139 t/s
Median across providers
Latency
402ms
Time to first token
Cost per task
$0.06
Estimated
Benchmarks
Where the score comes from
The Intelligence Index is a composite. These are the underlying evaluations this model was actually measured on.
Evaluation scores
Percentage correct · higher is better
- τ²-bench (Telecom)86.3%
- GPQA Diamond72.4%
- IFBench71.2%
- AA-LCR (long context)32.3%
- SciCode25.5%
- Humanity's Last Exam10.3%
- Terminal-Bench Hard7.6%
An evaluation missing from this list was not run for this model — it is not a zero.
View as table
| Evaluation | Score |
|---|---|
| τ²-bench (Telecom) | 86.3% |
| GPQA Diamond | 72.4% |
| IFBench | 71.2% |
| AA-LCR (long context) | 32.3% |
| SciCode | 25.5% |
| Humanity's Last Exam | 10.3% |
| Terminal-Bench Hard | 7.6% |
Against its peers
Intelligence Index · this model highlighted, nearest peers in grey
- 7.9
- Solar Pro 3Upstage7.8
- GPT-4.1 NanoOpenAI7.8
- SonarPerplexity7.7
- 7.7
- GPT-4o (2024-08-06)OpenAI7.7
- Sonar ProPerplexity7.6
- 7.3
Peers are the models sitting closest on the Intelligence Index — the set you would realistically choose between.
View as table
| Model | Intelligence |
|---|---|
| Qwen3 VL 30B A3B Instruct | 7.9 |
| Solar Pro 3 | 7.8 |
| GPT-4.1 Nano | 7.8 |
| Sonar | 7.7 |
| Qwen2.5 72B Instruct | 7.7 |
| GPT-4o (2024-08-06) | 7.7 |
| Sonar Pro | 7.6 |
| Qwen3 VL 8B Instruct | 7.3 |
Percentile among all indexed models
Pricing
What it costs to run
List prices per million tokens, plus what one representative task works out to.
List price
- Input / 1M tokens
- $0.15
- Output / 1M tokens
- $0.6
- Cached input / 1M
- $0.015
- Blended 3:1
- $0.262
One task, estimated
$0.06
- Input tokens
- 50,000
- Output tokens
- 80,000
- Profile
- Reasoning
Estimated from list pricing: 50K input tokens plus 80K output tokens for reasoning models (25K for non-reasoning).
Serving providers
Speed and latency figures are medians across these providers, so a widely-served model reports a blend rather than any single endpoint.