- Models
- 7
- listed and live
- Location
- Non-EU
- where requests are served
- Trains on prompts
- Not declared
- nothing published yet
- Keeps prompts
- Not declared
- nothing published yet
Models
What Novita serves on llm.buzz, at its own prices.
| Model | Input | Output | Cache read | Context | Latency | Uptime (3d) |
|---|---|---|---|---|---|---|
 google/gemma-4-26b-a4b-it | $0.13 | $0.40 | n/a | 256K | 0.83 s | 100% |
 moonshotai/kimi-k3 | $3.00 | $15.00 | $0.30 | 1M | 2.98 s | 100% |
 openai/gpt-oss-120b | $0.05 | $0.25 | n/a | 128K | 0.58 s | 94.4% |
 qwen/qwen3.8-27b | $0.42 | $3.00 | $0.085 | 256K | 2.48 s | 90.3% |
qwen/qwen3.8-flash | $0.15 | $0.47 | $0.016 | 1M | 1.20 s | 100% |
 z-ai/glm-5.3 | $1.40 | $4.40 | $0.26 | 1M | 1.72 s | 100% |
z-ai/glm-5.3-flash | $0.15 | $0.50 | $0.03 | 1M | 2.98 s | 100% |
Novita's own prices, in USD per million tokens. Latency is the median time to first token over 24 hours.
Call Novita's models with one keyThe same key reaches every provider here; each request pays the price of the provider that answered it.