Upstage: Solar Pro 4

Solar Pro 4 is Upstage's cost-efficient large language model, featuring a 524K context window. It is built for long-horizon tasks and agentic workflows, with strong capabilities in office productivit…

upstage/solar-pro4-20260810

Solar Pro 4 is Upstage's cost-efficient large language model, featuring a 524K context window. It is built for long-horizon tasks and agentic workflows, with strong capabilities in office productivity, document-intensive work, and coding.

When using BYOK with ZDR enforcement, the Console API key must belong to a ZDR-enabled Upstage organization. Contact Upstage to enable ZDR on your org.

Specifications & capabilities

Input
text
Output
text
Context window · tokens
524288
Capabilities
promptCaching, reasoning, structuredOutputs, toolCalling
Description source ↗

Evaluations

1–20 / 97

OpenRouter Benchmarks

80.808%

source id
openRouterBenchmarks
leaderboard name
GPQA Diamond · Accuracy · KT
dimension
knowledge
metric
GPQA Diamond · Accuracy
rank
4
sample size
198
aggregation scope
status
published
harness
url
https://github.com/OpenRouterTeam/benchmark-harness
name
OpenRouter benchmark harness
direction
higherIsBetter
run_count
1
conditions
window
newest 90 days; successful visible nonzero runs
dataset
benchmark-leaderboard
benchmark
id
gpqa_diamond
name
GPQA Diamond
record_kind
aggregate
provider_name
KT
attribution
Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
last_run_at
2026-11-11T02:02:05.405Z
license_url
https://creativecommons.org/licenses/by/4.0/
methodology
Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
metric_kind
GPQA Diamond · Accuracy
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
sample_size
198
source_file
benchmark-leaderboard.json
source_page
https://openrouter.ai/benchmarks/gpqa-diamond
measurements
accuracy
0.808081
cost_per_task_usd
0.05294089393939394
output_tokens_per_task
43927.25252525252
latency_seconds_per_task
409.68816161616166
published_at
2026-11-16T01:46:02.653Z
snapshot_date
2026-11-16
source_revision
7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
transformations
TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
metric_population
4
benchmark_evaluation
true
openrouter_permaslug
upstage/solar-pro4-20260810
source model id
upstage/solar-pro4-20260810
Original result ↗

OpenRouter Benchmarks

66%

source id
openRouterBenchmarks
leaderboard name
tau2-bench Verified Airline · Accuracy · KT
dimension
agentic
metric
tau2-bench Verified Airline · Accuracy
rank
4
sample size
50
aggregation scope
status
published
harness
url
https://github.com/OpenRouterTeam/benchmark-harness
name
OpenRouter benchmark harness
direction
higherIsBetter
run_count
1
conditions
window
newest 90 days; successful visible nonzero runs
dataset
benchmark-leaderboard
benchmark
id
tau_bench_verified_airline
name
tau2-bench Verified Airline
record_kind
aggregate
provider_name
KT
attribution
Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
last_run_at
2026-11-11T02:34:15.627Z
license_url
https://creativecommons.org/licenses/by/4.0/
methodology
Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
metric_kind
tau2-bench Verified Airline · Accuracy
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
sample_size
50
source_file
benchmark-leaderboard.json
source_page
https://openrouter.ai/benchmarks/tau2-bench-airline
measurements
accuracy
0.66
cost_per_task_usd
0.0747839
output_tokens_per_task
14969.76
latency_seconds_per_task
223.46761999999998
published_at
2026-11-16T01:46:02.653Z
snapshot_date
2026-11-16
source_revision
7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
transformations
TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
metric_population
4
benchmark_evaluation
true
openrouter_permaslug
upstage/solar-pro4-20260810
source model id
upstage/solar-pro4-20260810
Original result ↗

LMArena

-0.256

source id
lmarena
leaderboard name
LMArena Agent Bash Recovery Steps Overall
dimension
agentic
metric
Arena score
rank
52
sample size
23430
evaluated at
2026-10-08T00:00:00.000Z
source model id
Solar Pro 4
confidence lower
-0.30070562107506127
confidence upper
-0.2104427362891036
Original result ↗

LMArena

-0.035

source id
lmarena
leaderboard name
LMArena Agent Tool Hallucination Overall
dimension
agentic
metric
Arena score
rank
54
sample size
9770
evaluated at
2026-11-11T00:00:00.000Z
source model id
Solar Pro 4
confidence lower
-0.05102443149443374
confidence upper
-0.018329433352625563
Original result ↗

LMArena

-0.208

source id
lmarena
leaderboard name
LMArena Agent Steerability Overall
dimension
agentic
metric
Arena score
rank
57
sample size
6238
evaluated at
2026-11-11T00:00:00.000Z
source model id
Solar Pro 4
confidence lower
-0.24852889970854306
confidence upper
-0.16729638833797192
Original result ↗

LMArena

-0.242

source id
lmarena
leaderboard name
LMArena Agent Praise Complaint Overall
dimension
agentic
metric
Arena score
rank
58
sample size
2068
evaluated at
2026-11-11T00:00:00.000Z
source model id
Solar Pro 4
confidence lower
-0.28284166741310085
confidence upper
-0.20132380159842428
Original result ↗

LMArena

-0.226

source id
lmarena
leaderboard name
LMArena Agent Overall
dimension
agentic
metric
Arena score
rank
59
sample size
11044
evaluated at
2026-11-11T00:00:00.000Z
source model id
Solar Pro 4
confidence lower
-0.24784766627839938
confidence upper
-0.20408062787276932
Original result ↗

LMArena

-0.299

source id
lmarena
leaderboard name
LMArena Agent Task Outcome Explicit Overall
dimension
agentic
metric
Arena score
rank
59
sample size
3784
evaluated at
2026-11-11T00:00:00.000Z
source model id
Solar Pro 4
confidence lower
-0.35852348462757494
confidence upper
-0.2403704609747565
Original result ↗

LMArena

-0.465

source id
lmarena
leaderboard name
LMArena Agent Bash Reliability Overall
dimension
general
metric
Arena score
rank
59
sample size
399
evaluated at
2026-11-11T00:00:00.000Z
source model id
Solar Pro 4
confidence lower
-0.5558329205902771
confidence upper
-0.37504459175537974
Original result ↗

LMArena

1,434.873

source id
lmarena
leaderboard name
LMArena Text Factuality Industry Mathematical
dimension
factuality
metric
Arena rating
rank
88
sample size
1344
evaluated at
2026-11-09T00:00:00.000Z
source model id
Solar Pro 4
confidence lower
1419.3063629211806
confidence upper
1450.4392377402041
Original result ↗

LMArena

1,411.624

source id
lmarena
leaderboard name
LMArena Text Factuality Spanish
dimension
factuality
metric
Arena rating
rank
89
sample size
690
evaluated at
2026-11-09T00:00:00.000Z
source model id
Solar Pro 4
confidence lower
1391.458307635177
confidence upper
1431.7895087198672
Original result ↗

LMArena

1,459.405

source id
lmarena
leaderboard name
LMArena Text Factuality Expert
dimension
factuality
metric
Arena rating
rank
94
sample size
2695
evaluated at
2026-11-09T00:00:00.000Z
source model id
Solar Pro 4
confidence lower
1447.8600325702246
confidence upper
1470.9496773651927
Original result ↗

LMArena

1,382.679

source id
lmarena
leaderboard name
LMArena Text Korean
dimension
general
metric
Arena rating
rank
98
sample size
448
evaluated at
2026-11-09T00:00:00.000Z
source model id
Solar Pro 4
confidence lower
1354.8612313555675
confidence upper
1410.4961053019344
Original result ↗

LMArena

1,421.973

source id
lmarena
leaderboard name
LMArena Text Factuality Math
dimension
factuality
metric
Arena rating
rank
99
sample size
1142
evaluated at
2026-11-09T00:00:00.000Z
source model id
Solar Pro 4
confidence lower
1404.6674266519346
confidence upper
1439.2789604875068
Original result ↗

LMArena

1,461.079

source id
lmarena
leaderboard name
LMArena Text Factuality Chinese
dimension
factuality
metric
Arena rating
rank
101
sample size
1846
evaluated at
2026-11-09T00:00:00.000Z
source model id
Solar Pro 4
confidence lower
1446.9868958351685
confidence upper
1475.171298282205
Original result ↗

LMArena

1,360.717

source id
lmarena
leaderboard name
LMArena Webdev Webdev React
dimension
coding
metric
Arena rating
rank
113
sample size
2855
evaluated at
2026-11-10T00:00:00.000Z
source model id
Solar Pro 4
confidence lower
1347.5131877105055
confidence upper
1373.9204473869554
Original result ↗

LMArena

1,375.157

source id
lmarena
leaderboard name
LMArena Text Style Control Korean
dimension
style
metric
Arena rating
rank
114
sample size
448
evaluated at
2026-11-09T00:00:00.000Z
source model id
Solar Pro 4
confidence lower
1346.8627529576274
confidence upper
1403.4504379284867
Original result ↗

LMArena

1,371.068

source id
lmarena
leaderboard name
LMArena Webdev Webdev Html
dimension
coding
metric
Arena rating
rank
116
sample size
785
evaluated at
2026-11-10T00:00:00.000Z
source model id
Solar Pro 4
confidence lower
1347.1767885695726
confidence upper
1394.9587168697783
Original result ↗

LMArena

1,427.573

source id
lmarena
leaderboard name
LMArena Text Factuality Industry Legal And Government
dimension
factuality
metric
Arena rating
rank
118
sample size
1955
evaluated at
2026-11-09T00:00:00.000Z
source model id
Solar Pro 4
confidence lower
1413.9831851507856
confidence upper
1441.1634239767714
Original result ↗

LMArena

1,369.109

source id
lmarena
leaderboard name
LMArena Webdev Overall
dimension
coding
metric
Arena rating
rank
118
sample size
4501
evaluated at
2026-11-10T00:00:00.000Z
source model id
Solar Pro 4
confidence lower
1358.6262190664236
confidence upper
1379.5912274068558
Original result ↗

Published prices

1–3 / 3

upstage/solar-pro4 ↗

$0.03

Inference provider
openRouter
Platform
OpenRouter
Unit
cache_read_tokens_per_million
Currency
USD
Billing conditions

upstage/solar-pro4 ↗

$0.15

Inference provider
openRouter
Platform
OpenRouter
Unit
input_tokens_per_million
Currency
USD
Billing conditions

upstage/solar-pro4 ↗

$0.60

Inference provider
openRouter
Platform
OpenRouter
Unit
output_tokens_per_million
Currency
USD
Billing conditions

Source specification observations

1–2 / 2

openRouter ↗

field
context_length
value
524288
normalized value
524288
scope
source-model
source id
openRouter
confidence
1

openRouter ↗

field
max_output_tokens
value
131072
normalized value
131072
scope
source-model
source id
openRouter
confidence
1

Original sources

1–4 / 4

OpenRouter · Model page ↗

LMArena · Catalog ↗

OpenRouter · Catalog ↗

OpenRouter · Catalog ↗

Source links

1–4 / 4

Explore more models

Models

Loading
EVIDENCE / COMPARISON

Benchmarks

Explore every benchmark

Loading

These charts contain model-linked results. The charts above also include unlinked source profiles. Compare scores within each evaluation. Model quality, usage, transparency and system performance measure different things.

Install TensorTrace