# Novita AI API pricing and models

Novita AI publishes per-token prices for 93 text models, as of October 20, 2026. The lowest output rate is $0.02 per million tokens, on Llama 3.2 1B Instruct, and the highest-scoring model it serves is Qwen3.8 Max, at 81 on our 100-point score.

Novita AI runs a serverless API for open-weight models and lists its prices in a public model catalog.

Source: [Novita AI pricing page](https://novita.ai/pricing), read 2026-10-20. Prices in US dollars per million tokens.

## Models we score

| Model | Score | Input | Output | Cached input | Context | vs maker |
|---|---|---|---|---|---|---|
| Qwen3.8 Max (`qwen/qwen3.8-max`) | 81 | $2.00 | $6.00 | $0.25 | 1M | — |
| Kimi K3 (`moonshotai/kimi-k3`) | 79 | $3.00 | $15.00 | $0.30 | 1.05M | Same |
| Kimi K3 (`moonshotai/kimi-k3-p`) | 79 | $3.00 | $15.00 | $0.30 | 1.05M | Same |
| Step 5 Preview (`stepfun/step-5-preview`) | 77 | $1.00 | $2.70 | $0.10 | 1M | Same |
| Qwen3.7 Max (`qwen/qwen3.7-max`) | 76 | $1.25 | $3.75 | $0.25 | 1M | — |
| Ling 3.1 Flash (`inclusionai/ling-3.1-flash`) | 74 | $0.30 | $0.90 | $0.06 | 262K | — |
| MiMo V2.6 Pro (`xiaomimimo/mimo-v2.6-pro`) | 73 | $0.43 | $0.87 | $0.004 | 1.05M | Same |
| DeepSeek V4.1 Flash (`deepseek/deepseek-v4.1-flash`) | 73 | $0.30 | $1.20 | $0.006 | 1.05M | Same |
| DeepSeek V4.1 Flash (`deepseek/deepseek-v4.1-flash-p`) | 73 | $0.30 | $1.20 | $0.006 | 1.05M | Same |
| DeepSeek V4.1 Flash 1017 (`deepseek/deepseek-v4.1-flash-1017`) | 73 | $0.30 | $1.20 | $0.006 | 1.05M | Same |
| Qwen3.8 27B (`qwen/qwen3.8-27b`) | 73 | $0.42 | $3.00 | $0.085 | 1M | — |
| Hy4 Preview (`tencent/hy4-preview`) | 72 | $0.83 | $2.50 | $0.042 | 1M | — |
| MiMo V2.6 Flash (`xiaomimimo/mimo-v2.6-flash`) | 69 | $0.14 | $0.28 | $0.003 | 1.05M | Same |
| GLM 5.3 Flash (`zai-org/glm-5.3-flash`) | 69 | $0.15 | $0.50 | $0.03 | 1.05M | — |
| DeepSeek V4 Pro 0813 (`deepseek/deepseek-v4-pro-0813`) | 69 | $1.32 | $3.96 | $0.044 | 1.05M | Same |
| Xiaomi MiMo V2.5 Pro (`xiaomimimo/mimo-v2.5-pro`) | 68 | $0.52 | $1.04 | $0.004 | 1.05M | — |
| GLM 5.3 (`zai-org/glm-5.3`) | 68 | $1.40 | $4.40 | $0.26 | 1.05M | — |
| Hy3 (`tencent/hy3`) | 67 | $0.14 | $0.58 | $0.035 | 262K | — |
| Kimi K2.6 (`moonshotai/kimi-k2.6`) | 67 | $0.80 | $3.40 | $0.16 | 262K | -15% |
| Ling 3.0 Flash VL (`inclusionai/ling-3.0-flash-vl`) | 66 | $0.075 | $0.22 | $0.015 | 262K | — |
| MiniMax M3 (`minimax/minimax-m3`) | 65 | $0.30 | $1.20 | $0.06 | 1M | Same |
| GLM 5.2 (`zai-org/glm-5.2`) | 65 | $1.40 | $4.40 | $0.26 | 1.05M | Same |
| GLM 5.1 (`zai-org/glm-5.1`) | 64 | $1.38 | $4.40 | $0.26 | 205K | Same |
| Kimi K2.7 Code (`moonshotai/kimi-k2.7-code`) | 63 | $0.95 | $4.00 | $0.19 | 262K | Same |
| Xiaomi MiMo V2.5 (`xiaomimimo/mimo-v2.5`) | 62 | $0.17 | $0.34 | $0.003 | 1.05M | — |
| Step 3.7 Flash (`stepfun/step-3.7-flash`) | 62 | $0.20 | $1.15 | $0.04 | 262K | Same |
| DeepSeek V4 Flash 0731 (`deepseek/deepseek-v4-flash-0731`) | 62 | $0.44 | $1.32 | $0.028 | 1.05M | +371% |
| GLM 5 (`zai-org/glm-5`) | 62 | $1.00 | $3.20 | $0.20 | 203K | Same |
| Qwen3.6 27B (`qwen/qwen3.6-27b`) | 62 | $0.60 | $3.60 | — | 262K | — |
| Llama 3.3 70B Instruct (`meta-llama/llama-3.3-70b-instruct`) | 61 | $0.14 | $0.40 | — | 12K | — |
| Qwen3.5 122B A10B (`qwen/qwen3.5-122b-a10b`) | 61 | $0.40 | $3.20 | — | 262K | — |
| Ling 3.0 Flash (`inclusionai/ling-3.0-flash`) | 60 | $0.06 | $0.18 | $0.012 | 262K | — |
| MiniMax M2.7 (`minimax/minimax-m2.7`) | 60 | $0.30 | $1.20 | $0.06 | 205K | Same |
| DeepSeek R1 0528 (`deepseek/deepseek-r1-0528`) | 60 | $0.70 | $2.50 | $0.35 | 164K | — |
| Kimi K2.5 (`moonshotai/kimi-k2.5`) | 60 | $0.60 | $3.00 | $0.10 | 262K | Same |
| Qwen3.5 27B (`qwen/qwen3.5-27b`) | 59 | $0.30 | $2.40 | — | 262K | — |
| Gemma 4 26B A4B (`google/gemma-4-26b-a4b-it`) | 58 | $0.13 | $0.40 | — | 262K | — |
| MiniMax M1 (`minimaxai/minimax-m1-80k`) | 58 | $0.55 | $2.20 | — | 1M | — |
| Gemma 4 31B (`google/gemma-4-31b-it`) | 56 | $0.14 | $0.40 | — | 262K | — |
| GLM 4.7 Flash (`zai-org/glm-4.7-flash`) | 56 | $0.07 | $0.40 | $0.01 | 200K | — |
| Llama 4 Maverick Instruct (`meta-llama/llama-4-maverick-17b-128e-instruct-fp8`) | 56 | $0.27 | $0.85 | — | 1.05M | — |
| OpenAI GPT OSS 120B (`openai/gpt-oss-120b`) | 55 | $0.05 | $0.25 | — | 131K | — |
| Qwen3.6 35B A3B (`qwen/qwen3.6-35b-a3b`) | 55 | $0.25 | $1.49 | — | 262K | — |
| Qwen3.5 35B A3B (`qwen/qwen3.5-35b-a3b`) | 55 | $0.25 | $2.00 | — | 262K | — |
| GLM 4.5 Air (`zai-org/glm-4.5-air`) | 53 | $0.13 | $0.85 | $0.025 | 131K | -23% |
| GLM 4.6 (`zai-org/glm-4.6`) | 53 | $0.55 | $2.20 | $0.11 | 205K | — |
| GLM 4.7 (`zai-org/glm-4.7`) | 53 | $0.60 | $2.20 | $0.11 | 205K | — |
| GLM 4.7 (`zai-org/glm-4.7-h`) | 53 | $0.60 | $2.20 | $0.11 | 205K | — |
| DeepSeek R1 Turbo (`deepseek/deepseek-r1-turbo`) | 53 | $0.70 | $2.50 | — | 64K | +14% |
| DeepSeek R1 (`deepseek/deepseek-r1/community`) | 53 | $4.00 | $4.00 | — | 64K | +83% |
| Kimi K2 Instruct (`moonshotai/kimi-k2-instruct`) | 52 | $0.57 | $2.30 | — | 131K | -8% |
| Llama 3.1 8B Instruct (`meta-llama/llama-3.1-8b-instruct`) | 51 | $0.02 | $0.05 | — | 16K | — |
| Llama 4 Scout Instruct (`meta-llama/llama-4-scout-17b-16e-instruct`) | 47 | $0.18 | $0.59 | — | 131K | — |
| DeepSeek V3 (`deepseek/deepseek_v3`) | 43 | $0.89 | $0.89 | — | 64K | -19% |
| Gemma 3 27B (`google/gemma-3-27b-it`) | 41 | $0.12 | $0.20 | — | 98K | — |
| Ling 3.0 Flash Fin (`inclusionai/ling-3.0-flash-fin`) | 0 | $0.075 | $0.22 | $0.015 | 262K | — |
| Qwen3 Omni 30B A3B Thinking (`qwen/qwen3-omni-30b-a3b-thinking`) | 0 | $0.25 | $0.97 | — | 66K | — |
| MiniMax M2.5 (`minimax/minimax-m2.5`) | 0 | $0.30 | $1.20 | $0.03 | 205K | Same |

## Other models

| Model | Input | Output | Cached input | Context |
|---|---|---|---|---|
| Llama 3.2 1B Instruct (`meta-llama/llama-3.2-1b-instruct`) | $0.02 | $0.02 | — | 131K |
| PaddleOCR VL (`paddlepaddle/paddleocr-vl`) | $0.02 | $0.02 | — | 16K |
| DeepSeek OCR 2 (`deepseek/deepseek-ocr-2`) | $0.03 | $0.03 | — | 8K |
| Llama 3 8B Stheno V3.2 (`Sao10K/L3-8B-Stheno-v3.2`) | $0.05 | $0.05 | — | 8K |
| Baichuan M2 32B (`baichuan/baichuan-m2-32b`) | $0.07 | $0.07 | — | 131K |
| DeepSeek R1 Distill Qwen3 8B 0528 (`deepseek/deepseek-r1-0528-qwen3-8b`) | $0.06 | $0.09 | — | 128K |
| MythoMax L2 13B (`gryphe/mythomax-l2-13b`) | $0.09 | $0.09 | — | 4K |
| Gemma 3 12B (`google/gemma-3-12b-it`) | $0.05 | $0.10 | — | 131K |
| AutoGLM Phone 9B Multilingual (`zai-org/autoglm-phone-9b-multilingual`) | $0.035 | $0.14 | — | 66K |
| Hermes 2 Pro Llama 3 8B (`nousresearch/hermes-2-pro-llama-3-8b`) | $0.14 | $0.14 | — | 8K |
| NVIDIA Nemotron 3 Nano 30B A3B (`nvidia/nemotron-3-nano-30b-a3b`) | $0.05 | $0.20 | — | 262K |
| Ling 3.0 Flash Sante (`inclusionai/ling-3.0-flash-sante`) | $0.075 | $0.22 | $0.015 | 262K |
| ERNIE 4.5 21B A3B (`baidu/ernie-4.5-21B-a3b`) | $0.07 | $0.28 | — | 120K |
| Qwen3.8 Flash (`qwen/qwen3.8-flash`) | $0.15 | $0.47 | $0.016 | 1M |
| WizardLM 2 8x22B (`microsoft/wizardlm-2-8x22b`) | $0.62 | $0.62 | — | 66K |
| Qwen3 VL 30B A3B Instruct (`qwen/qwen3-vl-30b-a3b-instruct`) | $0.20 | $0.70 | — | 131K |
| Qwen MT Plus (`qwen/qwen-mt-plus`) | $0.25 | $0.75 | — | 16K |
| GLM 4.6V (`zai-org/glm-4.6v`) | $0.30 | $0.90 | $0.055 | 131K |
| ming image 0.1 design (`ming-image-0.1-design`) | $1.00 | $1.00 | — | — |
| ming image 0.1 design layer (`ming-image-0.1-design-layer`) | $1.00 | $1.00 | — | — |
| CoBuddy (`baidu/cobuddy`) | $0.28 | $1.13 | $0.07 | 131K |
| DeepSeek V4.1 Flash DST (`deepseek/deepseek-v4.1-flash-dst`) | $0.30 | $1.20 | $0.006 | 1.05M |
| deepseek/deepseek-v4-flash-0731-p (`deepseek/deepseek-v4-flash-0731-p`) | $0.44 | $1.32 | $0.015 | 1.05M |
| Macaron V1 Tall (`mindai/macaron-v1-tall`) | $0.25 | $1.43 | $0.044 | 262K |
| Llama 3 70B Euryale V2.1 (`sao10k/l3-70b-euryale-v2.1`) | $1.48 | $1.48 | — | 8K |
| GLM 4.5V (`zai-org/glm-4.5v`) | $0.60 | $1.80 | $0.11 | 66K |
| MiniMax M2.7-highspeed (`minimax/minimax-m2.7-highspeed`) | $0.60 | $2.40 | $0.06 | 205K |
| Macaron V1 Venti (`mindai/macaron-v1-venti`) | $0.82 | $2.48 | $0.17 | 1.05M |
| Kimi K2 Instruct 0905 (`moonshotai/kimi-k2-0905`) | $0.60 | $2.50 | — | 262K |
| Kimi K2 Thinking (`moonshotai/kimi-k2-thinking`) | $0.60 | $2.50 | $0.15 | 262K |
| DeepSeek V4 Pro (`deepseek/deepseek-v4-pro`) | $1.60 | $3.20 | $0.14 | 1.05M |
| Qwen3.5 397B A17B (`qwen/qwen3.5-397b-a17b`) | $0.60 | $3.60 | — | 262K |
| deepseek/deepseek-v4-pro-0813-p (`deepseek/deepseek-v4-pro-0813-p`) | $1.32 | $3.96 | $0.044 | 1.05M |
| zai-org/glm-5.3-p (`zai-org/glm-5.3-p`) | $1.40 | $4.40 | $0.26 | 1.05M |
| Qwen3.8 2.4T A95B (`qwen/qwen3.8-2.4t-a95b`) | $2.00 | $6.00 | $0.25 | 1M |

## Billing notes

- Batch requests are discounted 50% on input and output for supported models, which Novita describes as an introductory offer.
- A few models bill by prompt length. Their rows show the first tier, with the other tiers in the row note.

## Hosts that sell the same models

- [OpenRouter](https://benchlm.ai/hosts/openrouter): 48 shared, 30 cheaper per output token (DeepSeek V4.1 Flash 1017: $0.12 vs $1.20)
- [DeepInfra](https://benchlm.ai/hosts/deepinfra): 32 shared, 24 cheaper per output token (DeepSeek V4 Flash 0731: $0.18 vs $1.32)
- [Nebius Token Factory](https://benchlm.ai/hosts/nebius): 14 shared, 1 cheaper per output token (DeepSeek V4 Flash 0731: $0.28 vs $1.32)
- [Together AI](https://benchlm.ai/hosts/together): 13 shared, 1 cheaper per output token (DeepSeek V4 Flash 0731: $0.28 vs $1.32)
- [Baseten](https://benchlm.ai/hosts/baseten): 8 shared, 1 cheaper per output token (DeepSeek V4 Flash 0731: $0.26 vs $1.32)
- [Fireworks AI](https://benchlm.ai/hosts/fireworks): 7 shared, 0 cheaper per output token

## FAQ

### How much does the Novita AI API cost?

Novita AI bills per million tokens, with separate input and output rates for each model. On October 20, 2026 its output rates ran from $0.02 on Llama 3.2 1B Instruct to $15.00 across 93 paid models. Input is usually cheaper than output, so a long prompt with a short reply costs less than the output rate suggests.

### Which models can you run on Novita AI?

93 text models are listed, and 53 of them match a model we score. The highest-scoring is Qwen3.8 Max at 81. The table on this page has every model with its input, output and cached-input rate, the context window, and the date we read the price.

### What are the alternatives to Novita AI?

OpenRouter lists 48 of the same models and charges less output per token on 30 of them. DeepInfra and Nebius Token Factory also overlap. Price is only part of the choice: speed, rate limits and which model snapshot a host serves differ too.

Canonical page: https://benchlm.ai/hosts/novita