← Model list

Gemini 3 Pro Preview

google·google/gemini-3-pro-preview·GA·Closed·gemini-pro series

Preview Gemini flagship for complex reasoning, coding, and rich multimodal prompts

At a glance
Reference price$4.00 / $18.00 per 1M
Cache read $0.20 · OrcaRouter
Lowest paid$0.57 / $3.43
QiHang Gateway · 7× spread
Context1,048,576
⚠ Providers report 1,000,000–1,048,576; the table below is authoritative
Output limit65,536
Capabilities
ReasoningTool useStructured outputTemperatureAttachments
⚠ Providers report capability flags inconsistently
Modalities
TextImageVideoAudioPDF
Knowledge cutoff2025-01
Released / updated2025-11-18 / 2025-11-18

Quality & performanceArtificial Analysis · Intelligence Index v4.1 · rep. tier high

Intelligence40.6
Value32
Coding
Agentic
Output speed
TTFT
Value formulaIQ 40.6 ÷ min blended $1.285 = 32
Reasoning tier → intelligence / speed (higher tier = stronger but slower)
lowIQ 33.9
highIQ 40.6

Quality is independently evaluated by Artificial Analysis. Speed/latency are model-level medians.

Available at 8 providers7 with public prices

ProviderTierInputOutputCache readCache writeContextOutput limitStatus
QiHangGateway$0.57$3.431,000,00065,000
Jiekou.AIGateway$1.80$10.801,048,57665,536
Databricks
databricks-gemini-3-pro
Cloud$2.00
tiered >200K: $4.00
$12.00$0.201,048,57665,536
Merge GatewayGateway$2.00
tiered >200K: $4.00
$12.00$0.201,048,57665,536
302.AIGateway$2.00$12.001,000,00064,000
NEAR AI Cloud
google/gemini-3-pro
Gateway$1.25$15.00$01,048,57665,536
OrcaRouterGateway$4.00
tiered >200K: $4.00
$18.00$0.201,048,57665,536
AnyAPIGateway1,048,57665,536hostTable.undisclosed

Sorted by blended price (input×0.75 + output×0.25) asc. The official channel always shows regardless of rank. Whether a gateway's low price is actually usable can't be verified.

Your usage cost

1QiHang$285.50
2Databricks$784.00
3Merge Gateway$784.00
4NEAR AI Cloud$850.00
5Jiekou.AI$900.00
6302.AI$1,000.00
The cheapest paid channel is the only channel.

Benchmark1 items

NameConditionsScoreMetricSource
SWE-Bench Prodataset: public43.3resolve rateSource ↗

The same benchmark scores very differently across harness / dataset, so the qualifying conditions must be shown together.

Reasoning control

effort = loweffort = mediumeffort = high

3 / 8 providers expose no reasoning control (reasoning_options: []).

Related models

Price historyone sample accumulated per data sync

Input listOutput listMin blended
$18.00$02026-08-052026-08-132026-08-05 · Input list $2.002026-08-06 · Input list $2.002026-08-07 · Input list $2.002026-08-08 · Input list $2.002026-08-09 · Input list $2.002026-08-10 · Input list $2.002026-08-11 · Input list $4.002026-08-12 · Input list $4.002026-08-13 · Input list $4.002026-08-05 · Output list $12.002026-08-06 · Output list $12.002026-08-07 · Output list $12.002026-08-08 · Output list $12.002026-08-09 · Output list $12.002026-08-10 · Output list $12.002026-08-11 · Output list $18.002026-08-12 · Output list $18.002026-08-13 · Output list $18.002026-08-05 · Min blended $1.282026-08-06 · Min blended $1.282026-08-07 · Min blended $1.282026-08-08 · Min blended $1.282026-08-09 · Min blended $1.282026-08-10 · Min blended $1.282026-08-11 · Min blended $1.282026-08-12 · Min blended $1.282026-08-13 · Min blended $1.28