DeepSeek V4.1 Flash 1017 vs Gemini Nano Banana 2.1

Compare up to 4 models side by side.


Capability
Capability	159	—
Human preference	—	—
Priceper 1M tokens
Input	$0.015–$0.33list $0.30-95%↓	$1.35list $1.50-10%↑ ×90.0
Output	$0.06–$1.32list $1.20-95%↓	$6.75list $7.50-10%↑ ×112.5
Cache read	$0.0003–$0.028list $0.006-95%↓	$0.135–$1.35list $0.15-10%↑ ×450.0
Reasoning output	—	$6.75list $7.50-10%
Image input	—	$1.35list $1.50-10%
Image output	—	$27.00list $30.00-10%
Web searchper request	$0.000732↓	$0.0126list $0.014-10%↑ ×17.2
Limits and modalities
Context	1M↑	131K↓ 87%
Max output	384K↑	33K↓ 91%
Modalities	InputTextImageOutputText	InputTextImageVideoPDFOutputTextImage
Availability
Providers	ByteDance (Volcano Engine), Fireworks AI, Mistral AI, Tencent Cloud TokenHub, Together AI, DeepSeek, Alibaba Cloud (Model Studio Intl), Cloudflare Workers AI	Google (Gemini), Google Vertex AI
30-day uptime	—	—
Measured, last 30 days
Avg latency	—	—
Effective price	—	—
Cache hit rate	—	—
About
Released	2026-10-17	2026-10-06
Knowledge cutoff	—	—

Capability scores are from Epoch AI’s Epoch Capabilities Index (CC BY 4.0, as of 2026-11-02), which combines dozens of benchmarks with GPT-5 set at 150. epoch.ai/eci