[TRANSLATE]

[TRANSLATE]

[TRANSLATE]

[TRANSLATE]

• [TRANSLATE]
• [TRANSLATE]
• [TRANSLATE]

[TRANSLATE]

[TRANSLATE]

[TRANSLATE]

• [TRANSLATE]
• [TRANSLATE]
• [TRANSLATE]
• [TRANSLATE]

[TRANSLATE]

[TRANSLATE]

• [TRANSLATE]
• [TRANSLATE]
• [TRANSLATE]
• [TRANSLATE]
• [TRANSLATE]

[TRANSLATE]

[TRANSLATE]

GPT-5.4	Gemini 3.1 Flash-Lite	DeepSeek V4
[TRANSLATE]	OpenAI	Google	[TRANSLATE]
[TRANSLATE]	[TRANSLATE]	[TRANSLATE]	[TRANSLATE]
Context Window	1M tokens	1M tokens (64K output)	1M tokens
Parameters	[TRANSLATE]	[TRANSLATE]	1T total / 32B active (MoE)
Multimodal	Text, Image, Computer-use	Text, Image, Audio	Text, Image, Video, Audio
Open-source	ไม่	ไม่	[TRANSLATE]
[TRANSLATE]	[TRANSLATE]	[TRANSLATE]	[TRANSLATE]
[TRANSLATE]	Standard / Thinking / Pro	[TRANSLATE]	[TRANSLATE]

[TRANSLATE]

[TRANSLATE]

Input ($/1M tokens)	Output ($/1M tokens)	[TRANSLATE]
GPT-5.4 Standard	~$5.00-$15.00	~$15.00-$60.00	- (Baseline)
Gemini 3.1 Flash-Lite	$0.25	$1.50	20-60x
DeepSeek V4	$0.10-$0.30	~$0.50-$1.00	15-50x

[TRANSLATE]

[TRANSLATE]

[TRANSLATE]

[TRANSLATE]	[TRANSLATE]
[TRANSLATE]	GPT-5.4 Thinking	[TRANSLATE]
[TRANSLATE]	GPT-5.4 (Codex)	[TRANSLATE]
[TRANSLATE]	Gemini 3.1 Flash-Lite	[TRANSLATE]
Content Moderation	Gemini 3.1 Flash-Lite	[TRANSLATE]
[TRANSLATE]	DeepSeek V4	[TRANSLATE]
Self-hosted / On-premise	DeepSeek V4	[TRANSLATE]
[TRANSLATE]	GPT-5.4	[TRANSLATE]

[TRANSLATE]

[TRANSLATE]

[TRANSLATE]

[TRANSLATE]

• [TRANSLATE]
• [TRANSLATE]
• [TRANSLATE]
• [TRANSLATE]

[TRANSLATE]

[TRANSLATE]

[TRANSLATE]

[TRANSLATE]	[TRANSLATE]
GPT-5.4	[TRANSLATE]	[TRANSLATE]
Gemini 3.1 Flash-Lite	[TRANSLATE]	[TRANSLATE]
DeepSeek V4	[TRANSLATE]	[TRANSLATE]

[TRANSLATE]

[TRANSLATE]

[TRANSLATE]

• OpenAI — Introducing GPT-5.4
• Google — Gemini 3.1 Flash-Lite
• DeepSeek — DeepSeek V4 Technical Report