Decart

Data handling can vary by model and deployment. The privacy badge above shows the more conservative of the provider-wide fallback and the policies for routes on this page; each model page shows the policy for that specific route.

Models served by Decart

Model name

9 models

Model	Context	Input /M	Output /M	Cache read /M	Cache hit	TTFT	Throughput
DeepSeek V4.1 Flash FP4 deepseek/deepseek-v4.1-flash:decart	1.0M	$0.09/M	$0.19/M	$0.02/M	81.1%	0.9s	88 tps
DeepSeek V4.1 Flash (Thinking) FP4 deepseek/deepseek-v4.1-flash:thinking:decart	1.0M	$0.09/M	$0.19/M	$0.02/M	81.1%	0.9s	88 tps
DeepSeek V4.1 Flash 1017 FP4 deepseek/deepseek-v4.1-flash-1017:decart	1.0M	$0.11/M	$0.43/M	$0.0022/M	30.4%	1s	56 tps
DeepSeek V4.1 Flash 1017 (Thinking) FP4 deepseek/deepseek-v4.1-flash-1017:thinking:decart	1.0M	$0.11/M	$0.43/M	$0.0022/M	30.4%	1s	56 tps
GLM 5.2 FP4 z-ai/glm-5.2:decart	1.0M	$1.26/M	$8.40/M	$0.50/M	0%	0.4s	581 tps
GLM 5.2 (Thinking) FP4 z-ai/glm-5.2:thinking:decart	1.0M	$1.26/M	$8.40/M	$0.50/M	0%	0.4s	581 tps
GLM 5.3 FP4 z-ai/glm-5.3:decart	1.0M	$0.88/M	$2.82/M	$0.20/M	0.4%	1.2s	68 tps
GLM 5.3 (Thinking) FP4 z-ai/glm-5.3:thinking:decart	1.0M	$0.88/M	$2.82/M	$0.20/M	0%	1.2s	68 tps
Kimi K3 MXFP4 moonshotai/kimi-k3:decart	1.0M	$1.73/M	$8.66/M	$0.17/M	33.3%	1.3s	55 tps

DeepSeek V4.1 Flash

FP4

deepseek/deepseek-v4.1-flash:decart

Context

1.0M

Input /M

$0.09/M

Output /M

$0.19/M

Cache read /M

$0.02/M

Cache hit

81.1%

TTFT

0.9s

Throughput

88 tps

DeepSeek V4.1 Flash (Thinking)

FP4

deepseek/deepseek-v4.1-flash:thinking:decart

Context

1.0M

Input /M

$0.09/M

Output /M

$0.19/M

Cache read /M

$0.02/M

Cache hit

81.1%

TTFT

0.9s

Throughput

88 tps

DeepSeek V4.1 Flash 1017

FP4

deepseek/deepseek-v4.1-flash-1017:decart

Context

1.0M

Input /M

$0.11/M

Output /M

$0.43/M

Cache read /M

$0.0022/M

Cache hit

30.4%

TTFT

1s

Throughput

56 tps

DeepSeek V4.1 Flash 1017 (Thinking)

FP4

deepseek/deepseek-v4.1-flash-1017:thinking:decart

Context

1.0M

Input /M

$0.11/M

Output /M

$0.43/M

Cache read /M

$0.0022/M

Cache hit

30.4%

TTFT

1s

Throughput

56 tps

GLM 5.2

FP4

z-ai/glm-5.2:decart

Context

1.0M

Input /M

$1.26/M

Output /M

$8.40/M

Cache read /M

$0.50/M

Cache hit

0%

TTFT

0.4s

Throughput

581 tps

GLM 5.2 (Thinking)

FP4

z-ai/glm-5.2:thinking:decart

Context

1.0M

Input /M

$1.26/M

Output /M

$8.40/M

Cache read /M

$0.50/M

Cache hit

0%

TTFT

0.4s

Throughput

581 tps

GLM 5.3

FP4

z-ai/glm-5.3:decart

Context

1.0M

Input /M

$0.88/M

Output /M

$2.82/M

Cache read /M

$0.20/M

Cache hit

0.4%

TTFT

1.2s

Throughput

68 tps

GLM 5.3 (Thinking)

FP4

z-ai/glm-5.3:thinking:decart

Context

1.0M

Input /M

$0.88/M

Output /M

$2.82/M

Cache read /M

$0.20/M

Cache hit

0%

TTFT

1.2s

Throughput

68 tps

Kimi K3

MXFP4

moonshotai/kimi-k3:decart

Context

1.0M

Input /M

$1.73/M

Output /M

$8.66/M

Cache read /M

$0.17/M

Cache hit

33.3%

TTFT

1.3s

Throughput

55 tps

Prices are per million tokens. TTFT is the observed time to first token. Selectable routes include the applicable provider-selection markup; fixed routes show the standard model price. Availability and pricing refresh continuously; each model page shows the live provider comparison.

Using Decart via the API

Call a model served by Decart using its displayed model ID, including the provider suffix when selection is supported.

curl https://nano-gpt.com/api/v1/chat/completions \
  -H "Authorization: Bearer $NANOGPT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek/deepseek-v4.1-flash:decart",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

For models with provider selection, append :decart to the model ID, set "provider": "decart" in the request body, or send an X-Provider: decart header. Explicit provider selection adds a route-specific markup over that provider's base price; the prices in the table above already include it. Models without provider selection use the displayed model ID without a provider suffix or selection surcharge.

See the API documentation for provider preferences, price-aware routing, and error behavior.