Skip to main content

Deep Infra

Inference host

Deep Infra runs open and licensed models on its own hardware and sells them by usage, often faster or cheaper than the maker.

PricingDocsStatus
Visit deepinfra.com
Models88
Cheapest for
24 models
Serves
2 model types
Visits in September 2026502K−8.1%Unique visitors 170K
  1. Apr
  2. May
  3. Jun
  4. Jul
  5. Aug
  6. Sep
Models by type

Models on Deep Infra

ModelInputOutputvs officialContext
Qwen3.5 27B$0.26/1MFrom$2.60/1M
from $0.677 at AIHubMix
+8%—
Gemma 3 12B IT$0.05/1MFrom$0.15/1M
from $0.10 at Helicone
+50%—
Nemotron 3 Super 120B A12BCheapest$0.085/1MFrom$0.40/1M——
Nemotron 3 Ultra 550B A55BCheapest$0.50/1MFrom$2.20/1M——
Claude Haiku 4.5 (latest)$1/1M$5/1M
from $4 at APIMart
±0%—
Claude Opus 4.7$5/1M$25/1M
from $20 at APIMart
±0%—
Claude Opus 4.8$5/1M$25/1M
from $20 at APIMart
±0%—
Deepseek 3.2$0.26/1MFrom$0.38/1M
from $0.308 at APIMart
+23%—
DeepSeek R1 0528$0.50/1MFrom$2.15/1M
from $1.23 at APIMart
+74%—
DeepSeek V4 Flash Vision Exp$0.216/1MFrom$0.647/1M
from $0.48 at CometAPI
+35%—
DeepSeek V4 Pro$1.30/1MFrom$2.60/1M
from $1.58 at CometAPI
+31%—
DeepSeek-V3.1$0.25/1MFrom$0.95/1M
from $0.798 at Poe
+19%—
Doubao Seed 2.0 Code Preview$0.50/1M$3/1M
from $2.24 at ZenMux
±0%—
Doubao Seed 2.0 Mini 260428$0.10/1M$0.40/1M
from $0.10 at Fal
±0%—
Doubao Seed 2.0 Pro$0.50/1M$3/1M
from $2 at CometAPI
±0%—
Doubao-Seed-1.8$0.25/1M$2/1M
from $0.28 at ZenMux
+614%—
Gemini 3.1 Flash Lite$0.25/1M$1.50/1M
from $1.20 at CometAPI
±0%—
Gemini 3.5 Flash$1.50/1M$9/1M
from $2.70 at Kie.ai
+233%—
Gemini 3.7 Flash$0.75/1M$3.75/1M
from $1.13 at Kie.ai
+233%—
Granite 4.2 8BCheapest$0.06/1M$0.25/1M±0%—
Hy4 preview$0.834/1M$2.50/1M
from $2.25 at OpenRouter
+11%—
Inkling SmallCheapest$0.45/1M$1.20/1M±0%—
SL3-8B-Lunaris-v1Cheapest$0.04/1MFrom$0.05/1M——
SL31 70B Euryale V2.2Cheapest$0.85/1MFrom$0.85/1M——
Ling 3.0 Flash VLCheapest$0.06/1M$0.18/1M±0%—
Llama Guard 4 12BCheapest$0.18/1MFrom$0.18/1M±0%—
Mistral Small 3Cheapest$0.05/1MFrom$0.08/1M——
Mistral Small 3.2 24BCheapest$0.075/1MFrom$0.20/1M±0%—
Muse Glimmer 30B$0.30/1M$1.20/1M
from $0.80 at EmpirioLabs AI
+50%—
Nemotron 3.5 Super VL 120B A12B$0.15/1M$0.50/1M——
Nemotron-Content-Safety-3.5Only here$0.20/1MFrom$0.20/1M——
Phi 4Cheapest$0.07/1MFrom$0.14/1M±0%—
Qwen2.5-72B-InstructCheapest$0.36/1MFrom$0.40/1M±0%—
Qwen3 235B A22B Instruct 2507$0.09/1MFrom$0.55/1M
from $0.10 at Requesty
+450%—
Qwen3 Next 80B A3B InstructCheapest$0.09/1MFrom$1.10/1M±0%—
Qwen3 VL 235B A22B InstructCheapest$0.20/1MFrom$0.88/1M±0%—
Qwen3 VL 30B A3B Instruct$0.15/1MFrom$0.60/1M
from $0.389 at SiliconFlow
+54%—
Step 5 PreviewCheapest$1/1M$2.70/1M±0%—

Deep Infra at a glance

88 models in total: 86 text, 2 moderation. The cheapest is Mistral Nemo at $0.03 /1M. Llama 4 Scout 17B Instruct has the largest context window, 1.31M tokens. The most searched is Kimi K3, with about 658K Google searches a month. The newest is DeepSeek V4 Flash Vision Exp, released Oct 17, 2026.

Cheapest

  1. Mistral Nemo$0.03/1M
  2. SL3-8B-Lunaris-v1$0.05/1M
  3. Mistral Small 3$0.08/1M
  4. Gemma 3 4B IT$0.10/1M
  5. GPT OSS 20B$0.14/1M

Largest context

  1. Llama 4 Scout 17B Instruct1.31M
  2. Kimi K31.05M
  3. GLM-5.31.05M
  4. DeepSeek V4.1 Flash1.05M
  5. GLM-5.3-Flash1.05M

Most searched

  1. Kimi K3658K
  2. Claude Opus 5.5405K
  3. GLM-5.3276K
  4. DeepSeek V4.1 Flash138K
  5. GLM-5.3-Flash121K

Newest

  1. DeepSeek V4 Flash Vision ExpOct 17, 2026
  2. Nemotron 3.5 Super VL 120B A12BOct 15, 2026
  3. Ling 3.1 FlashOct 13, 2026
  4. Claude Sonnet 5.5Sep 28, 2026
  5. Claude Opus 5.5Sep 22, 2026