Akash

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 Akash

Model name

10 models

Model	Context	Input /M	Output /M	Cache read /M	Cache hit	TTFT	Throughput
GLM 5.3 FP8 z-ai/glm-5.3:akash	1.0M	$0.20/M	$4.62/M	$0.20/M	0%	1s	72 tps
GLM 5.3 (Thinking) FP8 z-ai/glm-5.3:thinking:akash	1.0M	$0.20/M	$4.62/M	$0.20/M	0%	1s	72 tps
GPT OSS 120B BF16 openai/gpt-oss-120b:akash	131k	$0.03/M	$0.20/M	$0.03/M	N/A	0.7s	37 tps
GPT OSS 20B FP4 openai/gpt-oss-20b:akash	131k	$0.02/M	$0.10/M	N/A	0%	1s	26 tps
Kimi K3 FP4 moonshotai/kimi-k3:akash	1.0M	$1.26/M	$14.70/M	$1.26/M	79%	1.2s	65 tps
Llama 3.3 70B Instruct FP8 meta-llama/llama-3.3-70b-instruct:akash	131k	$0.21/M	$0.55/M	$0.10/M	8.2%	1s	40 tps
Qwen 3.6 35B A3B FP8 Qwen/Qwen3.6-35B-A3B:akash	262k	$0.10/M	$0.94/M	$0.05/M	N/A	0.5s	56 tps
Qwen 3.6 35B A3B Thinking FP8 Qwen/Qwen3.6-35B-A3B:thinking:akash	262k	$0.10/M	$0.94/M	$0.05/M	N/A	0.5s	56 tps
Qwen3.8 27B FP8 qwen3.8-27b:akash	262k	$0.24/M	$2.08/M	$0.05/M	N/A	0.7s	43 tps
Qwen3.8 27B Thinking FP8 qwen3.8-27b:thinking:akash	262k	$0.24/M	$2.08/M	$0.05/M	N/A	0.7s	43 tps

GLM 5.3

FP8

z-ai/glm-5.3:akash

Context

1.0M

Input /M

$0.20/M

Output /M

$4.62/M

Cache read /M

$0.20/M

Cache hit

0%

TTFT

1s

Throughput

72 tps

GLM 5.3 (Thinking)

FP8

z-ai/glm-5.3:thinking:akash

Context

1.0M

Input /M

$0.20/M

Output /M

$4.62/M

Cache read /M

$0.20/M

Cache hit

0%

TTFT

1s

Throughput

72 tps

GPT OSS 120B

BF16

openai/gpt-oss-120b:akash

Context

131k

Input /M

$0.03/M

Output /M

$0.20/M

Cache read /M

$0.03/M

Cache hit

N/A

TTFT

0.7s

Throughput

37 tps

GPT OSS 20B

FP4

openai/gpt-oss-20b:akash

Context

131k

Input /M

$0.02/M

Output /M

$0.10/M

Cache read /M

N/A

Cache hit

0%

TTFT

1s

Throughput

26 tps

Kimi K3

FP4

moonshotai/kimi-k3:akash

Context

1.0M

Input /M

$1.26/M

Output /M

$14.70/M

Cache read /M

$1.26/M

Cache hit

79%

TTFT

1.2s

Throughput

65 tps

Llama 3.3 70B Instruct

FP8

meta-llama/llama-3.3-70b-instruct:akash

Context

131k

Input /M

$0.21/M

Output /M

$0.55/M

Cache read /M

$0.10/M

Cache hit

8.2%

TTFT

1s

Throughput

40 tps

Qwen 3.6 35B A3B

FP8

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

Context

262k

Input /M

$0.10/M

Output /M

$0.94/M

Cache read /M

$0.05/M

Cache hit

N/A

TTFT

0.5s

Throughput

56 tps

Qwen 3.6 35B A3B Thinking

FP8

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

Context

262k

Input /M

$0.10/M

Output /M

$0.94/M

Cache read /M

$0.05/M

Cache hit

N/A

TTFT

0.5s

Throughput

56 tps

Qwen3.8 27B

FP8

qwen3.8-27b:akash

Context

262k

Input /M

$0.24/M

Output /M

$2.08/M

Cache read /M

$0.05/M

Cache hit

N/A

TTFT

0.7s

Throughput

43 tps

Qwen3.8 27B Thinking

FP8

qwen3.8-27b:thinking:akash

Context

262k

Input /M

$0.24/M

Output /M

$2.08/M

Cache read /M

$0.05/M

Cache hit

N/A

TTFT

0.7s

Throughput

43 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 Akash via the API

Call a model served by Akash 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": "z-ai/glm-5.3:akash",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

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