Novita AI vs Together AI
Same workloads, both price lists, refreshed daily. On shared line items today: Novita AI is cheaper on 10 of 15 shared models (input or output price), Together AI on 1.
Novita AI
Combines a serverless LLM API (DeepSeek, Llama, Qwen at aggressive per-token prices) with GPU instances and template deployments, one of the few providers covering both sides of the hosting equation..
- Both LLM API and GPU rental under one account
- Aggressive open-weight model pricing
- Template marketplace for common stacks
Together AI
High-performance open-weight inference (Llama, DeepSeek, Qwen) on a custom stack, plus fine-tuning and GPU clusters. Consistently among the fastest and cheapest for open models.
- Top-tier open-model throughput
- Fine-tuning pipeline
- Dedicated endpoints and clusters
Shared models, priced by both
List prices refreshed 2026-08-23 ยท input $/1M tokens| Model | Novita AI in/out | Together AI in/out | Cheaper input |
|---|---|---|---|
| GPT-OSS 20B | $0.040 / $0.150 | $0.050 / $0.200 | Novita AI |
| GPT-OSS 120B | $0.050 / $0.250 | $0.150 / $0.600 | Novita AI |
| GLM-4.5 Air | $0.130 / $0.850 | $0.200 / $1.10 | Novita AI |
| Qwen3 Next 80B A3B Instruct | $0.150 / $1.50 | $0.150 / $1.50 | tie |
| Qwen3 Next 80B A3B Thinking | $0.150 / $1.50 | $0.150 / $1.50 | tie |
| Llama 4 Scout | $0.180 / $0.590 | $0.180 / $0.590 | tie |
| DeepSeek V3 | $0.270 / $1.12 | $1.25 / $1.25 | Novita AI |
| Llama 4 Maverick | $0.270 / $0.850 | $0.270 / $0.850 | tie |
| DeepSeek-V3.1 | $0.270 / $1.00 | $0.600 / $1.70 | Novita AI |
| Qwen3-Coder-480b-A35b-Instruct | $0.300 / $1.30 | $2.00 / $2.00 | Novita AI |
| Qwen3 235B A22B Thinking 2507 | $0.300 / $3.00 | $0.650 / $3.00 | Novita AI |
| GLM-4.6 | $0.550 / $2.20 | $0.600 / $2.20 | Novita AI |
| Kimi K2 | $0.600 / $2.50 | $1.00 / $3.00 | Novita AI |
| GLM-4.7 | $0.600 / $2.20 | $0.450 / $2.00 | Together AI |
| DeepSeek R1 | $0.700 / $2.50 | $3.00 / $7.00 | Novita AI |
Cheaper-on-count is a tally of listed prices, not a quality verdict: throughput, reliability and quantization differ between providers. Disclosure.