Solar Open2 250B
Aug 2026
Solar Open2 250BCompare inon
Benchmarks of Solar Open2 250B12
BenchmarkByScoreRank
Overall1
Knowledge3
Science3
Agents2
Documents1
Legacy · saturated or no longer reported2
No benchmark matches
—per task?
self-host onlyprice
—output speed?
1Mcontext
Ranking?
Alternatives?
ModelScoreCostPer taskWhat changes
BetterGPT-6 Lunamedium+21%—$0.017Function calling only with none · Chat Completions+3
OpenStep 5 Previewthinking+77%—$1.03Memory (4-bit): 150 GB → 363 GB+3
BetterClaude Haiku 5.5low+19%—$0.024Remove sampling parameters+5BetterGPT-5.6 Lunahigh+30%—$0.044
BetterMiMo v2.6 FlashthinkingOpen+53%—$0.062Memory (4-bit): 150 GB → 186 GB+4ModelScoreCostPer taskWhat changes
Nothing here yet.
ModelScoreCostPer taskWhat changes
GPT-6 Lunamedium+21%—$0.017Function calling only with none · Chat Completions+3Claude Haiku 5.5low+19%—$0.024Remove sampling parameters+5GPT-5.6 Lunahigh+30%—$0.044MiMo v2.6 FlashthinkingOpen+53%—$0.062Memory (4-bit): 150 GB → 186 GB+4GPT-6.1 Sollow+70%—$0.13Remove unsupported parameters+5ModelScoreCostPer taskWhat changes
Step 5 PreviewthinkingOpen+77%—$1.03Memory (4-bit): 150 GB → 363 GB+3
GLM 5.3 FlashmaxOpen+69%—$0.28Memory (4-bit): 150 GB → 193 GB+5
Ling 3.1 FlashOpen+66%—$1.00Memory (4-bit): 150 GB → 343 GB
Qwen3.8 FlashthinkingOpen+61%—$0.37Context: 1M → 262K+6MiMo v2.6 FlashthinkingOpen+53%—$0.062Memory (4-bit): 150 GB → 186 GB+4ModelScoreCostPer taskWhat changes
Solar Pro 4thinking+7%—$0.71Context: 1M → 524KSolar Mini 4same—$0.37Context: 1M → 524KSolar Open 100BOpen-58%——Context: 1M → 128KSolar Pro 2 (Preview)high-63%——Context: 1M → 64KSolar Pro 3high-68%—$0.079Context: 1M → 131KSolar Pro 2high-70%—$0.15 / $0.60Context: 1M → 66KAll Upstage models →Run it yourself
estimateFits on✗ 8 GB (phone, laptop)✗ 16 GB✗ 24 GB GPU✗ 48 GB GPU✗ 80 GB GPU
Licenceupstage-solar-licensecommercial use with conditions
Benchmarks
all categoriesChat & Knowledge
Humanity's Last Exam28.5#87
AA Omniscience Index-1.8#76
CritPt (physics)5.7#75
AA Intelligence Index 4.3.2≈24.7#86
AA Omniscience · Accuracy18.8#169
AA-LCR (long context) 1.171.7#111
GPQA Diamond85.7#71
Coding
SciCode48#59
Terminal-Bench 2.144.2#78
Agents & Tool Use
GDPval-AA (work tasks) 2.123.5#85
τ³-Banking · AA21.6#50
FAQ
How much does Solar Open2 250B cost?›
There is no per-unit API price: self-host only.
Is Solar Open2 250B free to use?›
There is no free API. The weights are open, so you can run it yourself.
How good is Solar Open2 250B, and what is it best at?›
#87 of 334 in Chat & Knowledge, #59 of 140 in Coding, #85 of 175 in Agents & Tool Use. Strongest at: Science, Knowledge.
Is Solar Open2 250B better than Grok 4.3?›
On AA Intelligence Index 4.3.2, Solar Open2 250B scores 24.7 vs 24.9 — about the same.
What is the context window of Solar Open2 250B?›
1M tokens.
When was Solar Open2 250B released? Is there a newer version?›
Released 2026-08-12. It is the current version in its line.
Is Solar Open2 250B open source?›
Yes, the weights are open (upstage-solar-license licence).
Can I run Solar Open2 250B locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can Solar Open2 250B read images?›
No, it accepts text only.
What is Solar Open2 250B?›
Solar Open2 250B is an open-weights language model from Upstage, released Aug 2026.
What are the best alternatives to Solar Open2 250B?›
GPT-6 Luna (better), Step 5 Preview (open weights).