Novita AI
Inference hostNovita AI runs open and licensed models on its own hardware and sells them by usage, often faster or cheaper than the maker.
Models83
- Cheapest for
- 5 models
- Serves
- 3 model types
Visits in September 2026303K−4.1%Unique visitors 122K
- Apr
- May
- Jun
- Jul
- Aug
- Sep
Models by type
Models on Novita AI
| Model | Input | Output | vs official | Context |
|---|---|---|---|---|
 Baichuan M2 32BBaichuanCheapest | $0.07/1M | $0.07/1M | — | — |
 CoBuddyBaiduOnly here | $0.28/1M | $1.13/1M | — | — |
 DeepSeek OCR 2DeepSeek | $0.03/1M | $0.03/1M | — | — |
| DeepSeek R1 0528DeepSeek | $0.70/1M | $2.50/1M from $1.23 at APIMart | +103% | — |
| DeepSeek R1 0528 Qwen3 8BDeepSeekOnly here | $0.06/1M | $0.09/1M | — | — |
| DeepSeek V4 ProDeepSeek | $1.32/1M | $3.96/1M from $1.58 at CometAPI | +100% | — |
| ERNIE 4.5 21B A3BBaidu | $0.07/1M | $0.28/1M | — | — |
 GLM-4.5VZhipu AI | $0.60/1M | $1.80/1M from $0.822 at AIHubMix | ±0% | — |
| Hermes 2 Pro Llama 3 8BNous Research | $0.14/1M | $0.14/1M | — | — |
 Hy3Tencent | $0.14/1M | $0.58/1M from $0.33 at OpenRouter | +76% | — |
| Hy4 previewTencent | $0.834/1M | $2.50/1M from $2.25 at OpenRouter | +11% | — |
| Kimi K2 0905Moonshot AI | $0.60/1M | $2.50/1M from $2 at Helicone | +25% | — |
| Kimi K2 ThinkingMoonshot AI | $0.60/1M | $2.50/1M from $1.68 at APIMart | +49% | — |
| Kimi K3.1 Pro HighspeedMoonshot AI | $2/1M | From$10/1M from $6.13 at Kenari | +6% | — |
| SL3-70B-Euryale-v2.1Sao10K | $1.48/1M | $1.48/1M | — | — |
 Ling 3.0 Flash FinInclusionai | $0.075/1M | $0.22/1M from $0.18 at Hugging Face | +22% | — |
| Ling 3.0 Flash SanteInclusionaiCheapest | $0.075/1M | $0.22/1M | — | — |
| Ling 3.0 Flash VLInclusionai | $0.075/1M | $0.22/1M from $0.18 at Deep Infra | +22% | — |
 Macaron V1 TallMindlabCheapest | $0.247/1M | $1.43/1M | — | — |
| Macaron V1 VentiMindlabCheapest | $0.825/1M | $2.48/1M | — | — |
 MiniMax M1MiniMax | $0.55/1M | $2.20/1M | — | — |
| MiniMax-M2.7MiniMax | $0.30/1M | $1.20/1M from $0.60 at EmpirioLabs AI | +100% | — |
| GMythoMax L2 13BGryphe | $0.09/1M | $0.09/1M | — | — |
 PaddleOCR VLPaddlepaddleOnly here | $0.02/1M | $0.02/1M | — | — |
 Qwen MT PlusAlibaba | $0.25/1M | $0.75/1M | -3% | — |
| Qwen3 Omni 30B A3B ThinkingAlibaba | $0.25/1M | $0.97/1M from $0.389 at SiliconFlow | +149% | — |
| Qwen3 Reranker 8BAlibaba | $0.05/1M | $0.05/1M from $0.039 at SiliconFlow | +29% | — |
| Qwen3 VL 30B A3B InstructAlibaba | $0.20/1M | $0.70/1M from $0.389 at SiliconFlow | +80% | — |
| Qwen3.5 122B-A10BAlibaba | $0.40/1M | $3.20/1M from $0.901 at AIHubMix | ±0% | — |
| SSao10K Stheno 8bSao10K | $0.05/1M | $0.05/1M | — | — |
 Step 3.7 FlashStepFun | $0.20/1M | $1.15/1M from $0.925 at APIMart | +24% | — |
| Step 5 PreviewStepFun | $1/1M | $2.70/1M | ±0% | — |
 WizardLM-2 8x22BMicrosoft | $0.62/1M | $0.62/1M | — | — |
Novita AI at a glance
83 models in total: 80 text, 2 reranking, 1 embedding. The cheapest is BGE M3 at $0.01 /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 Kimi K3.1 Pro Highspeed, released Oct 23, 2026.
Cheapest
- BGE M3$0.01/1M
- BGE Reranker v2 M3$0.01/1M
- PaddleOCR VL$0.02/1M
- DeepSeek OCR 2$0.03/1M
Llama 3.1 8B Instruct$0.05/1M
Largest context
- Llama 4 Scout 17B Instruct1.31M
- Kimi K31.05M
- GLM-5.31.05M
- DeepSeek V4.1 Flash1.05M
- GLM-5.3-Flash1.05M
Most searched
- Kimi K3658K
- GLM-5.3276K
- Qwen3.8 2.4T A95B234K
- DeepSeek V4.1 Flash138K
- GLM-5.3-Flash121K
Newest
- Kimi K3.1 Pro HighspeedOct 23, 2026
- Ling 3.0 Flash SanteOct 19, 2026
- DeepSeek V4 Flash Vision ExpOct 17, 2026
- MiniMax-M3.1-FlashOct 17, 2026
- Ling 3.1 FlashOct 13, 2026