DeepInfra

69 models available

Models served by DeepInfra

Model	Context	Input /M	Output /M	Cache read /M	Cache hit	Latency	Throughput
DeepSeek R1-0528 FP4	128k	$0.53/M	$2.26/M	$0.37/M	59.6%	0.9s	22 tps
DeepSeek V3 0324 FP4	128k	$0.25/M	$0.94/M	$0.14/M	20.8%	1.1s	19 tps
DeepSeek V3.1 FP4	128k	$0.26/M	$1.00/M	$0.14/M	34%	0.7s	15 tps
DeepSeek V3.2 FP4	163k	$0.27/M	$0.40/M	$0.14/M	N/A	1.7s	13 tps
DeepSeek V3.2 (Thinking) FP4	163k	$0.27/M	$0.40/M	$0.14/M	N/A	1.7s	13 tps
DeepSeek V4 Flash FP8	1.0M	$0.09/M	$0.19/M	$0.02/M	0%	0.8s	24 tps
DeepSeek V4 Flash (Thinking) FP8	1.0M	$0.09/M	$0.19/M	$0.02/M	1%	0.8s	24 tps
DeepSeek V4 Flash 0731 FP8	1.0M	$0.08/M	$0.19/M	$0.02/M	19.3%	0.8s	29 tps
DeepSeek V4 Flash 0731 (Thinking) FP8	1.0M	$0.08/M	$0.19/M	$0.02/M	4%	0.8s	29 tps
DeepSeek V4 Flash Vision Exp FP8	1.0M	$0.23/M	$0.68/M	$0.07/M	75%	2.7s	35 tps
DeepSeek V4 Pro FP8	1.0M	$1.36/M	$2.73/M	$0.10/M	0%	1.1s	35 tps
DeepSeek V4 Pro (Thinking) FP8	1.0M	$1.36/M	$2.73/M	$0.10/M	80.9%	1.1s	35 tps
DeepSeek V4 Pro 0813 FP8	1.0M	$1.36/M	$2.73/M	$0.10/M	N/A	0.9s	69 tps
DeepSeek V4 Pro 0813 (Thinking) FP8	1.0M	$1.36/M	$2.73/M	$0.10/M	32.8%	0.9s	69 tps
Gemma 4 26B A4B FP8	262k	$0.07/M	$0.36/M	$0.01/M	0%	0.6s	24 tps
Gemma 4 31B FP8	262k	$0.14/M	$0.40/M	$0.02/M	3.7%	2.7s	61 tps
GLM 4.6 FP4	200k	$0.53/M	$2.10/M	$0.10/M	84.2%	0.6s	47 tps
GLM 4.6 (Thinking) FP4	200k	$0.53/M	$2.10/M	$0.10/M	75%	0.6s	47 tps
GLM 4.7 FP4	200k	$0.42/M	$1.84/M	$0.08/M	36.1%	0.9s	21 tps
GLM 4.7 (Thinking) FP4	200k	$0.42/M	$1.84/M	$0.08/M	28.5%	0.9s	21 tps
GLM 4.7 Flash (Thinking) BF16	200k	$0.06/M	$0.42/M	$0.01/M	N/A	0.4s	20 tps
GLM 4.7 Flash Original (Thinking) BF16	200k	$0.06/M	$0.42/M	$0.01/M	N/A	0.4s	20 tps
GLM 5 FP4	200k	$0.63/M	$2.18/M	$0.13/M	N/A	0.8s	30 tps
GLM 5 (Thinking) FP4	200k	$0.63/M	$2.18/M	$0.13/M	N/A	0.8s	30 tps
GLM 5.1 FP4	200k	$1.10/M	$3.68/M	$0.22/M	N/A	5.9s	8 tps
GLM 5.1 (Thinking) FP4	200k	$1.10/M	$3.68/M	$0.22/M	N/A	5.9s	8 tps
GLM 5.2 FP4	1.0M	$0.51/M	$1.64/M	$0.10/M	66.7%	1.2s	59 tps
GLM 5.2 (Thinking) FP4	1.0M	$0.51/M	$1.64/M	$0.10/M	66.7%	1.2s	59 tps
GLM 5.3 FP4	1.0M	$1.26/M	$4.20/M	$0.13/M	N/A	2s	45 tps
GLM 5.3 (Thinking) FP4	1.0M	$1.26/M	$4.20/M	$0.13/M	N/A	2s	45 tps
GLM 5.3 Flash FP4	1.0M	$0.08/M	$0.26/M	$0.02/M	7%	1.2s	34 tps
GPT OSS 120B FP8	128k	$0.21/M	$1.00/M	N/A	0%	1.9s	87 tps
GPT OSS 20B BF16	128k	$0.03/M	$0.15/M	N/A	N/A	0.4s	81 tps
Granite 4.2 8B BF16	131k	$0.06/M	$0.26/M	$0.02/M	73.4%	0.4s	110 tps
Hermes 3 70B FP8	66k	$0.73/M	$0.73/M	N/A	0%	0.8s	32 tps
Inkling Small FP8	524k	$0.47/M	$1.26/M	$0.10/M	N/A	0.9s	75 tps
Inkling Small Thinking FP8	524k	$0.47/M	$1.26/M	$0.10/M	N/A	0.9s	75 tps
Kimi K2.5 FP4	256k	$0.47/M	$2.36/M	$0.07/M	10.1%	0.8s	26 tps
Kimi K2.5 Thinking FP4	256k	$0.47/M	$2.36/M	$0.07/M	10.1%	0.8s	26 tps
Kimi K2.6 FP4	256k	$0.79/M	$3.68/M	$0.16/M	N/A	1.4s	28 tps
Kimi K2.7 Code FP4	262k	$0.71/M	$3.57/M	$0.14/M	0%	6.4s	42.5 tps
Kimi K3 BF16	1.0M	$2.99/M	$14.96/M	$0.30/M	17.2%	1.4s	22 tps
Llama 3.3 70B Instruct FP8	131k	$0.10/M	$0.34/M	N/A	0%	0.5s	16 tps
Llama 4 Maverick FP8	1.0M	$0.21/M	$0.84/M	N/A	N/A	0.4s	40 tps
MiMo V2.5 FP8	1.0M	$0.14/M	$0.68/M	$0.03/M	N/A	2.1s	25 tps
MiMo V2.5 Pro FP8	1.0M	$1.05/M	$3.15/M	$0.21/M	24.3%	0.7s	41 tps
MiMo V2.5 Pro Thinking FP8	1.0M	$1.05/M	$3.15/M	$0.21/M	0%	0.7s	41 tps
MiMo V2.5 Thinking FP8	1.0M	$0.14/M	$0.68/M	$0.03/M	N/A	2.1s	25 tps
MiniMax M2.7 FP8	205k	$0.26/M	$1.05/M	$0.05/M	1.4%	0.6s	27 tps
MiniMax M3 FP8	512k	$0.29/M	$1.16/M	$0.06/M	95.5%	0.9s	48 tps
MiniMax M3 Thinking FP8	512k	$0.29/M	$1.16/M	$0.06/M	0.4%	0.9s	48 tps
Nvidia Nemotron 3 Super 120B BF16	262k	$0.09/M	$0.42/M	N/A	0%	1.4s	54 tps
Nvidia Nemotron 3 Super 120B Thinking BF16	262k	$0.09/M	$0.42/M	N/A	0%	1.4s	54 tps
Nvidia Nemotron 3 Ultra 550B	256k	$0.53/M	$2.31/M	$0.11/M	38.3%	4.2s	3 tps
Nvidia Nemotron 3 Ultra 550B Thinking	256k	$0.53/M	$2.31/M	$0.11/M	0%	10.1s	307.4 tps
Nvidia Nemotron 3.5 Lightning	29k	$0.08/M	$0.21/M	$0.04/M	19.6%	0.9s	142.7 tps
Nvidia Nemotron 3.5 Lightning Thinking	29k	$0.08/M	$0.21/M	$0.04/M	N/A	N/A	N/A
Qwen 2.5 72B Instruct FP8	131k	$0.38/M	$0.42/M	N/A	0%	0.6s	15 tps
Qwen 3.6 27B FP8	260k	$0.34/M	$3.36/M	N/A	N/A	1.7s	23 tps
Qwen 3.6 27B Thinking FP8	260k	$0.34/M	$3.36/M	N/A	N/A	1.7s	23 tps
Qwen 3.6 35B A3B FP8	262k	$0.10/M	$1.00/M	N/A	N/A	1s	105 tps
Qwen 3.6 35B A3B Thinking FP8	262k	$0.10/M	$1.00/M	N/A	N/A	1s	105 tps
Qwen3 235B A22B FP8	262k	$0.09/M	$0.58/M	N/A	N/A	0.5s	13 tps
Qwen3 32B FP8	41k	$0.08/M	$0.29/M	N/A	0%	0.5s	25 tps
Qwen3 Next 80B A3B Instruct FP8	256k	$0.09/M	$1.16/M	N/A	0%	0.6s	71 tps
Qwen3.5 397B FP8	258k	$0.47/M	$3.15/M	$0.23/M	N/A	0.7s	35 tps
Qwen3.5 397B Thinking FP8	258k	$0.47/M	$3.15/M	$0.23/M	N/A	0.7s	35 tps
Qwen3.8 Max FP4	262k	$2.10/M	$6.30/M	$0.21/M	82.4%	0.6s	75.5 tps
Qwen3.8 Max Thinking FP4	262k	$2.10/M	$6.30/M	$0.21/M	82.4%	0.6s	75.5 tps

Prices are per million tokens and include the 5% provider-selection markup. Availability and pricing refresh continuously; each model page shows the live provider comparison.

Using DeepInfra via the API

Append :deepinfra to the model ID, set "provider": "deepinfra" in the request body, or send an X-Provider: deepinfra header. Explicit provider selection adds a 5% markup over that provider's base price; the prices in the table above already include it.

curl https://nano-gpt.com/api/v1/chat/completions \
  -H "Authorization: Bearer $NANOGPT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-ai/DeepSeek-R1-0528:deepinfra",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

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

DeepInfra terms of use DeepInfra privacy policy