DekaLLM

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 DekaLLM

Model name

11 models

Model	Context	Input /M	Output /M	Cache read /M	Cache hit	TTFT	Throughput
DeepSeek V4.1 Flash deepseek/deepseek-v4.1-flash:dekallm	1.0M	$0.13/M	$1.26/M	$0.0053/M	69.8%	1.5s	91 tps
DeepSeek V4.1 Flash (Thinking) deepseek/deepseek-v4.1-flash:thinking:dekallm	1.0M	$0.13/M	$1.26/M	$0.0053/M	86.6%	1.5s	91 tps
GLM 5.3 Flash z-ai/glm-5.3-flash:dekallm	1.0M	$0.10/M	$1.05/M	$0.04/M	N/A	2.7s	41 tps
GPT OSS 120B BF16 openai/gpt-oss-120b:dekallm	131k	$0.03/M	$0.19/M	$0.03/M	1.1%	0.7s	22 tps
GPT OSS 20B BF16 openai/gpt-oss-20b:dekallm	131k	$0.03/M	$0.15/M	$0.03/M	0.4%	0.7s	29 tps
Nvidia Nemotron 3 Super 120B FP8 nvidia/nemotron-3-super-120b-a12b:dekallm	262k	$0.08/M	$0.47/M	N/A	0%	1.8s	16 tps
Nvidia Nemotron 3 Super 120B Thinking FP8 nvidia/nemotron-3-super-120b-a12b:thinking:dekallm	262k	$0.08/M	$0.47/M	N/A	0%	1.8s	16 tps
Qwen 3.6 35B A3B Qwen/Qwen3.6-35B-A3B:dekallm	262k	$0.10/M	$1.05/M	$0.05/M	N/A	0.4s	37 tps
Qwen 3.6 35B A3B Thinking Qwen/Qwen3.6-35B-A3B:thinking:dekallm	262k	$0.10/M	$1.05/M	$0.05/M	N/A	0.4s	37 tps
Qwen3.8 27B qwen3.8-27b:dekallm	262k	$0.05/M	$3.15/M	$0.02/M	N/A	0.7s	107 tps
Qwen3.8 27B Thinking qwen3.8-27b:thinking:dekallm	262k	$0.05/M	$3.15/M	$0.02/M	N/A	0.7s	107 tps

DeepSeek V4.1 Flash

deepseek/deepseek-v4.1-flash:dekallm

Context

1.0M

Input /M

$0.13/M

Output /M

$1.26/M

Cache read /M

$0.0053/M

Cache hit

69.8%

TTFT

1.5s

Throughput

91 tps

DeepSeek V4.1 Flash (Thinking)

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

Context

1.0M

Input /M

$0.13/M

Output /M

$1.26/M

Cache read /M

$0.0053/M

Cache hit

86.6%

TTFT

1.5s

Throughput

91 tps

GLM 5.3 Flash

z-ai/glm-5.3-flash:dekallm

Context

1.0M

Input /M

$0.10/M

Output /M

$1.05/M

Cache read /M

$0.04/M

Cache hit

N/A

TTFT

2.7s

Throughput

41 tps

GPT OSS 120B

BF16

openai/gpt-oss-120b:dekallm

Context

131k

Input /M

$0.03/M

Output /M

$0.19/M

Cache read /M

$0.03/M

Cache hit

1.1%

TTFT

0.7s

Throughput

22 tps

GPT OSS 20B

BF16

openai/gpt-oss-20b:dekallm

Context

131k

Input /M

$0.03/M

Output /M

$0.15/M

Cache read /M

$0.03/M

Cache hit

0.4%

TTFT

0.7s

Throughput

29 tps

Nvidia Nemotron 3 Super 120B

FP8

nvidia/nemotron-3-super-120b-a12b:dekallm

Context

262k

Input /M

$0.08/M

Output /M

$0.47/M

Cache read /M

N/A

Cache hit

0%

TTFT

1.8s

Throughput

16 tps

Nvidia Nemotron 3 Super 120B Thinking

FP8

nvidia/nemotron-3-super-120b-a12b:thinking:dekallm

Context

262k

Input /M

$0.08/M

Output /M

$0.47/M

Cache read /M

N/A

Cache hit

0%

TTFT

1.8s

Throughput

16 tps

Qwen 3.6 35B A3B

Qwen/Qwen3.6-35B-A3B:dekallm

Context

262k

Input /M

$0.10/M

Output /M

$1.05/M

Cache read /M

$0.05/M

Cache hit

N/A

TTFT

0.4s

Throughput

37 tps

Qwen 3.6 35B A3B Thinking

Qwen/Qwen3.6-35B-A3B:thinking:dekallm

Context

262k

Input /M

$0.10/M

Output /M

$1.05/M

Cache read /M

$0.05/M

Cache hit

N/A

TTFT

0.4s

Throughput

37 tps

Qwen3.8 27B

qwen3.8-27b:dekallm

Context

262k

Input /M

$0.05/M

Output /M

$3.15/M

Cache read /M

$0.02/M

Cache hit

N/A

TTFT

0.7s

Throughput

107 tps

Qwen3.8 27B Thinking

qwen3.8-27b:thinking:dekallm

Context

262k

Input /M

$0.05/M

Output /M

$3.15/M

Cache read /M

$0.02/M

Cache hit

N/A

TTFT

0.7s

Throughput

107 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 DekaLLM via the API

Call a model served by DekaLLM 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:dekallm",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

For models with provider selection, append :dekallm to the model ID, set "provider": "dekallm" in the request body, or send an X-Provider: dekallm 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.