Novita: Models Intelligence, Performance & Price
Analysis of Novita's models across key metrics including quality, price, output speed, latency, context window & more. This analysis is intended to support you in choosing the best model provided by Novita for your use-case.
Most Intelligent
Intelligence index
Total 84 models
Fastest
Output speed
Total 84 models
Lowest Price
Blended price (per 1M tokens)
Total 84 models
Novita offers 84 models, each with different intelligence, performance, and pricing characteristics. Below is a comparison of the key metrics across models.
- For intelligence, the top models on Novita are GLM-5.2 (max) (FP8) (51), Qwen3.7 Max (46), and MiniMax-M3 (44).
- For output speed, the fastest models are NVIDIA Nemotron 3 Nano (FP4) (304 t/s), NVIDIA Nemotron 3 Nano (FP4) (275 t/s), and Qwen3 Next 80B A3B (175 t/s). Speed varies significantly across models, with a 91% difference between the fastest and slowest.
- For latency, Llama 4 Scout (0.83s), Llama 3.1 8B (0.89s), and Llama 4 Maverick (FP8) (0.92s) offer the lowest time to first answer token.
- For pricing, Llama 3.1 8B ($0.02), gpt-oss-20b (high) ($0.05), and gpt-oss-20b (low) ($0.05) offer the lowest blended prices per 1M tokens. Prices vary up to 2.8x across models.
- For context window size, GLM-5.2 (max) (FP8) (1M), DeepSeek V4 Pro (max) (1M), and DeepSeek V4 Pro (high) (1M) support the largest context windows on Novita.
Highlights
Intelligence Evaluations
Artificial Analysis Intelligence Index
Intelligence Evaluations
Agentic real-world work tasks, (Elo-500)/2000
Agentic tool use
Agentic coding & terminal use
Coding
Reasoning & knowledge
Scientific reasoning
Physics reasoning
Knowledge
1 - hallucination rate
Long context reasoning
Agentic knowledge work, Elo
Agentic SaaS workflows
Legal agentic work, criterion pass rate
Agentic business operations
Instruction following
Long-horizon agentic tasks
Kubernetes incident root-cause analysis
Visual reasoning
Intelligence Index vs. Price
Context Window
Context Window
Pricing
Intelligence Index vs. Price
Performance Summary
Output Speed vs. Price
Speed
Measured by Output Speed (tokens per second)
Output Speed
Latency
Measured by Time (seconds) to First Token
Latency: Time To First Answer Token
End-to-End Response Time
Seconds to output 500 tokens, calculated based on time to first token, 'thinking' time for reasoning models, and output speed
End-to-End Response Time vs. Price
Further Analysis | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
GLM-5.2 (max) (FP8) | 1.05M | Open | 51 | $0.42 | 49 | 2.20 | 53.54 | 41.07 | |||
Qwen3.7 Max | 1M | Proprietary | 46 | $0.65 | 51 | 2.03 | 58.80 | 47.01 | |||
MiniMax-M3 | 1M | Open | 44 | $0.13 | 88 | 1.25 | 29.82 | 22.85 | |||
DeepSeek V4 Pro (max) | 1.05M | Open | 44 | $0.23 | 59 | 1.66 | 83.74 | 73.66 | |||
Kimi K2.6 | 262k | Open | 44 | $0.33 | 53 | 1.60 | 95.02 | 83.99 | |||
DeepSeek V4 Pro (high) | 1.05M | Open | 43 | $0.20 | 58 | 1.80 | 44.84 | 34.41 | |||
 MiMo-V2.5-Pro | 1.05M | Open | 42 | $0.04 | 39 | 2.30 | 66.22 | 51.14 | |||
Kimi K2.7 Code | 262k | Open | 42 | $0.22 | 39 | 3.78 | 73.47 | 56.92 | |||
 Hy3 | 262k | Open | 41 | $0.04 | 76 | 2.71 | 35.73 | 26.41 | |||
DeepSeek V4 Flash (max) | 1.05M | Open | 40 | $0.06 | 109 | 1.74 | 57.85 | 51.53 | |||
GLM-5.1 (FP8) | 205k | Open | 40 | $0.30 | 49 | 1.94 | 90.29 | 78.05 | |||
GLM-5 FP8 | 203k | Open | 40* | -- | 40 | 2.01 | 93.21 | 78.55 | |||
MiniMax-M2.7 (FP8) | 205k | Open | 38 | $0.06 | 51 | 4.14 | 62.22 | 48.28 | |||
DeepSeek V4 Flash (high) | 1.05M | Open | 37 | $0.04 | 112 | 1.66 | 17.21 | 11.08 | |||
MiMo-V2.5 | 1.05M | Open | 37 | $0.01 | 53 | 2.02 | 49.10 | 37.66 | |||
Qwen3.6 27B | 262k | Open | 37 | $0.29 | 53 | 2.94 | 118.44 | 106.14 | |||
Kimi K2.5 | 262k | Open | 35 | $0.10 | 38 | 1.47 | 91.89 | 77.38 | |||
GLM-5.1 (FP8) | 205k | Open | 35* | -- | 55 | 2.45 | 11.47 | -- | |||
Kimi K2.6 | 262k | Open | 35* | -- | 45 | 1.75 | 12.79 | -- | |||
Qwen3.5 27B | 262k | Open | 34* | -- | 52 | 5.51 | 53.61 | 38.48 | |||
GLM-4.7 | 205k | Open | 34 | $0.35 | 40 | 3.77 | 65.70 | 49.55 | |||
 KAT-Coder-Pro V2 | 256k | Proprietary | 34 | -- | 105 | 1.46 | 6.22 | -- | |||
Qwen3.5 397B A17B | 262k | Open | 34 | $0.36 | 65 | 1.87 | 58.48 | 48.93 | |||
MiniMax-M2.5 | 205k | Open | 34* | -- | 128 | 1.82 | 21.35 | 15.62 | |||
Kimi K2 Thinking | 262k | Open | 33* | -- | 38 | 1.56 | 67.82 | 53.01 | |||
GLM-5 (FP8) | 203k | Open | 32* | -- | 40 | 2.03 | 14.65 | -- | |||
Qwen3.5 122B A10B | 262k | Open | 32 | $0.26 | 105 | 2.04 | 25.85 | 19.04 | |||
DeepSeek V3.2 | 164k | Open | 32 | $0.09 | 36 | 2.67 | 71.55 | 55.10 | |||
Qwen3.5 397B A17B | 262k | Open | 32* | -- | 61 | 1.94 | 10.08 | -- | |||
Qwen3.6 35B A3B | 205k | Open | 32 | $0.19 | 150 | 1.40 | 40.57 | 35.85 | |||
MiniMax-M2.1 | 205k | Open | 31* | -- | 80 | 1.83 | 33.15 | 25.06 | |||
Qwen3.6 27B | 262k | Open | 30 | $0.40 | 60 | 2.91 | 11.27 | -- | |||
DeepSeek V3.1 Terminus (FP8) | 131k | Open | 30 | -- | 35 | 2.88 | 73.88 | 56.80 | |||
Kimi K2.5 | 262k | Open | 29* | -- | 38 | 1.39 | 14.65 | -- | |||
 Gemma 4 31B | 262k | Open | 29 | $0.04 | -- | -- | -- | -- | |||
Qwen3.5 35B A3B | 262k | Open | 29* | -- | 134 | 1.56 | 20.28 | 14.97 | |||
GLM-4.6 | 205k | Open | 29 | $0.30 | 59 | 2.78 | 45.41 | 34.11 | |||
MiniMax-M2 | 205k | Open | 28* | -- | 90 | 1.65 | 29.35 | 22.16 | |||
MiMo-V2.5-Pro | 1.05M | Open | 28* | -- | 36 | 1.53 | 15.56 | -- | |||
GLM-4.7 | 205k | Open | 27* | -- | 39 | 3.66 | 16.35 | -- | |||
Gemma 4 26B A4B | 262k | Open | 26 | $0.04 | 32 | 3.04 | 82.09 | 63.24 | |||
DeepSeek V3.2 Exp (FP8) | 164k | Open | 25* | -- | 35 | 2.59 | 73.19 | 56.48 | |||
DeepSeek V3.2 | 164k | Open | 25* | -- | 37 | 2.72 | 16.34 | -- | |||
Qwen3.6 35B A3B | 205k | Open | 24 | $0.42 | 171 | 1.37 | 4.28 | -- | |||
Qwen3 Max | 262k | Proprietary | 24* | -- | 35 | 3.05 | 17.29 | -- | |||
gpt-oss-120b (high) | 131k | Open | 24 | $0.03 | 85 | 0.94 | 30.45 | 23.61 | |||
Kimi K2 0905 | 262k | Open | 24* | -- | 38 | 1.29 | 14.61 | -- | |||
GLM-4.6 | 205k | Open | 23* | -- | 60 | 2.64 | 10.95 | -- | |||
GLM-4.7-Flash | 200k | Open | 23* | -- | 92 | 1.71 | 28.91 | 21.76 | |||
Gemma 4 31B | 262k | Open | 22 | $0.04 | 44 | 1.51 | 12.82 | -- | |||
DeepSeek V3.1 Terminus (FP8) | 131k | Open | 21* | -- | 36 | 2.76 | 16.64 | -- | |||
DeepSeek V3.2 Exp (FP8) | 164k | Open | 21* | -- | 36 | 2.73 | 16.62 | -- | |||
Qwen3 Coder Next (FP8) | 262k | Open | 21 | $0.20 | 159 | 2.47 | 5.62 | -- | |||
DeepSeek V3.1 | 164k | Open | 21* | -- | 36 | 2.86 | 16.92 | -- | |||
DeepSeek V3.1 | 131k | Open | 21* | -- | 35 | 2.85 | 74.00 | 56.92 | |||
Qwen3 VL 235B A22B | 131k | Open | 21* | -- | 48 | 2.38 | 54.04 | 41.33 | |||
Gemma 4 26B A4B | 262k | Open | 20* | -- | 31 | 2.94 | 18.84 | -- | |||
DeepSeek R1 0528 | 164k | Open | 20* | -- | 28 | 1.29 | 91.72 | 72.34 | |||
Qwen3 235B A22B 2507 | 131k | Open | 20 | $0.08 | 70 | 2.25 | 37.93 | 28.54 | |||
Kimi K2 | 131k | Open | 19* | -- | 38 | 1.45 | 14.76 | -- | |||
DeepSeek R1 (Jan) Turbo | 64k | Open | 19 | $0.13 | -- | -- | -- | -- | |||
DeepSeek R1 (Jan) | 64k | Open | 19 | $0.57 | -- | -- | -- | -- | |||
Qwen3 235B 2507 | 131k | Open | 18* | -- | 44 | 1.69 | 13.08 | -- | |||
Qwen3 Coder 480B | 262k | Open | 18* | -- | 64 | 2.24 | 10.07 | -- | |||
MiniMax M1 80k | 1M | Open | 18* | -- | 77 | 2.00 | 34.37 | 25.90 | |||
GLM-4.6V | 131k | Open | 17* | -- | 79 | 3.60 | 35.22 | 25.29 | |||
GLM-4.7-Flash | 200k | Open | 16* | -- | 89 | 1.66 | 7.25 | -- | |||
DeepSeek V3 0324 | 164k | Open | 15 | $0.05 | 39 | 1.73 | 14.49 | -- | |||
gpt-oss-120b (low) | 131k | Open | 15 | $0.01 | 69 | 1.07 | 37.42 | 29.08 | |||
gpt-oss-20b (high) | 131k | Open | 15 | $0.02 | 72 | 1.13 | 36.09 | 27.97 | |||
gpt-oss-20b (low) | 131k | Open | 14* | -- | 74 | 1.18 | 35.03 | 27.08 | |||
Qwen3 VL 235B A22B | 131k | Open | 14* | -- | 37 | 2.08 | 15.66 | -- | |||
Llama 4 Maverick (FP8) | 1.05M | Open | 14 | $0.04 | 38 | 0.92 | 14.01 | -- | |||
NVIDIA Nemotron 3 Nano (FP4) | 262k | Open | 14 | $0.02 | 275 | 1.02 | 10.11 | 7.27 | |||
DeepSeek V3 (Dec) | 64k | Open | 14 | $0.06 | 40 | 1.70 | 14.31 | -- | |||
DeepSeek V3 (Dec) Turbo | 64k | Open | 14 | $0.03 | 39 | 1.72 | 14.44 | -- | |||
Ling 2.6 Flash | 262k | Open | 14 | -- | 84 | 1.10 | 7.03 | -- | |||
Qwen3 Next 80B A3B | 131k | Open | 14* | -- | 175 | 1.58 | 4.45 | -- | |||
GLM-4.6V | 131k | Open | 11* | -- | 87 | 3.67 | 9.44 | -- | |||
Qwen3 235B (FP8) | 41k | Open | 11* | -- | 28 | 1.37 | 19.45 | -- | |||
Llama 4 Scout | 131k | Open | 10 | $0.01 | 56 | 0.83 | 9.78 | -- | |||
Qwen3 VL 30B A3B | 131k | Open | 10* | -- | 83 | 1.56 | 7.57 | -- | |||
Llama 3.3 70B | 131k | Open | 9 | $0.02 | -- | -- | -- | -- | |||
GLM-4.5V | 65.5k | Open | 9* | -- | 84 | 1.76 | 31.47 | 23.77 | |||
Llama 3.1 8B | 16.4k | Open | 8 | -- | 146 | 0.89 | 4.33 | -- | |||
Gemma 3 27B | 98.3k | Open | 7 | $0.11 | 35 | 1.50 | 15.96 | -- | |||
NVIDIA Nemotron 3 Nano (FP4) | 262k | Open | 7* | -- | 304 | 1.02 | 2.67 | -- | |||
GLM-4.5V | 65.5k | Open | 7* | -- | 102 | 1.84 | 6.75 | -- | |||
Llama 3 70B | 8.19k | Open | 3* | -- | -- | -- | -- | -- | |||
Llama 3 8B | 8.19k | Open | 1* | -- | -- | -- | -- | -- | |||
Key definitions
Frequently Asked Questions
Common questions about Novita
Novita offers 84 models that we track: GLM-5.2 (max) (FP8), Qwen3.7 Max, MiniMax-M3, DeepSeek V4 Pro (max), Kimi K2.6, DeepSeek V4 Pro (high), MiMo-V2.5-Pro, Kimi K2.7 Code, Hy3, DeepSeek V4 Flash (max), GLM-5.1 (FP8), GLM-5 FP8, MiniMax-M2.7 (FP8), DeepSeek V4 Flash (high), MiMo-V2.5, Qwen3.6 27B, Kimi K2.5, GLM-5.1 (FP8), Kimi K2.6, Qwen3.5 27B, GLM-4.7, KAT-Coder-Pro V2, Qwen3.5 397B A17B, MiniMax-M2.5, Kimi K2 Thinking, GLM-5 (FP8), Qwen3.5 122B A10B, DeepSeek V3.2, Qwen3.5 397B A17B, Qwen3.6 35B A3B, MiniMax-M2.1, Qwen3.6 27B, DeepSeek V3.1 Terminus (FP8), Kimi K2.5, Qwen3.5 35B A3B, GLM-4.6, MiniMax-M2, MiMo-V2.5-Pro, GLM-4.7, Gemma 4 26B A4B, DeepSeek V3.2 Exp (FP8), DeepSeek V3.2, Qwen3.6 35B A3B, Qwen3 Max, gpt-oss-120b (high), Kimi K2 0905, GLM-4.6, GLM-4.7-Flash, Gemma 4 31B, DeepSeek V3.1 Terminus (FP8), DeepSeek V3.2 Exp (FP8), Qwen3 Coder Next (FP8), DeepSeek V3.1, DeepSeek V3.1, Qwen3 VL 235B A22B, Gemma 4 26B A4B, DeepSeek R1 0528, Qwen3 235B A22B 2507, Kimi K2, Qwen3 235B 2507, Qwen3 Coder 480B, MiniMax M1 80k, GLM-4.6V, GLM-4.7-Flash, DeepSeek V3 0324, gpt-oss-120b (low), gpt-oss-20b (high), gpt-oss-20b (low), Qwen3 VL 235B A22B, Llama 4 Maverick (FP8), NVIDIA Nemotron 3 Nano (FP4), DeepSeek V3 (Dec), DeepSeek V3 (Dec) Turbo, Ling 2.6 Flash, Qwen3 Next 80B A3B, GLM-4.6V, Qwen3 235B (FP8), Llama 4 Scout, Qwen3 VL 30B A3B, GLM-4.5V, Llama 3.1 8B, Gemma 3 27B, NVIDIA Nemotron 3 Nano (FP4), and GLM-4.5V.
The most intelligent model available on Novita is GLM-5.2 (max) (FP8) with an Intelligence Index score of 51.
The fastest model on Novita by output speed is NVIDIA Nemotron 3 Nano (FP4) at 303.6 tokens per second.
The model with the lowest time to first answer token on Novita is Llama 4 Scout at 0.83s. Lower latency means faster initial response time.
The most affordable model on Novita by blended price is Llama 3.1 8B at $0.02 per 1M tokens (7:2:1 cache hit/input/output ratio).
Prices on Novita vary up to 119x across models, from $0.02 per 1M tokens for Llama 3.1 8B to $2.74 per 1M tokens for Qwen3 Max.
Yes, Novita offers an OpenAI-compatible API, making it easy to switch from OpenAI or use existing OpenAI SDK integrations.
81 of 84 models on Novita support JSON mode for structured output.
79 of 84 models on Novita support function calling (tool use).
Yes, Novita offers 45 reasoning models: GLM-5.2 (max) (FP8), Qwen3.7 Max, MiniMax-M3, DeepSeek V4 Pro (max), Kimi K2.6, DeepSeek V4 Pro (high), MiMo-V2.5-Pro, Kimi K2.7 Code, Hy3, DeepSeek V4 Flash (max), GLM-5.1 (FP8), GLM-5 FP8, MiniMax-M2.7 (FP8), DeepSeek V4 Flash (high), MiMo-V2.5, Qwen3.6 27B, Kimi K2.5, Qwen3.5 27B, GLM-4.7, Qwen3.5 397B A17B, MiniMax-M2.5, Kimi K2 Thinking, Qwen3.5 122B A10B, DeepSeek V3.2, Qwen3.6 35B A3B, MiniMax-M2.1, DeepSeek V3.1 Terminus (FP8), Qwen3.5 35B A3B, GLM-4.6, MiniMax-M2, Gemma 4 26B A4B, DeepSeek V3.2 Exp (FP8), gpt-oss-120b (high), GLM-4.7-Flash, DeepSeek V3.1, Qwen3 VL 235B A22B, DeepSeek R1 0528, Qwen3 235B A22B 2507, MiniMax M1 80k, GLM-4.6V, gpt-oss-120b (low), gpt-oss-20b (high), gpt-oss-20b (low), NVIDIA Nemotron 3 Nano (FP4), and GLM-4.5V. Reasoning models use extended thinking to work through complex problems before providing an answer.
Yes, 81 of 84 models on Novita are open weight models: GLM-5.2 (max) (FP8), MiniMax-M3, DeepSeek V4 Pro (max), Kimi K2.6, DeepSeek V4 Pro (high), MiMo-V2.5-Pro, Kimi K2.7 Code, Hy3, DeepSeek V4 Flash (max), GLM-5.1 (FP8), GLM-5 FP8, MiniMax-M2.7 (FP8), DeepSeek V4 Flash (high), MiMo-V2.5, Qwen3.6 27B, Kimi K2.5, GLM-5.1 (FP8), Kimi K2.6, Qwen3.5 27B, GLM-4.7, Qwen3.5 397B A17B, MiniMax-M2.5, Kimi K2 Thinking, GLM-5 (FP8), Qwen3.5 122B A10B, DeepSeek V3.2, Qwen3.5 397B A17B, Qwen3.6 35B A3B, MiniMax-M2.1, Qwen3.6 27B, DeepSeek V3.1 Terminus (FP8), Kimi K2.5, Qwen3.5 35B A3B, GLM-4.6, MiniMax-M2, MiMo-V2.5-Pro, GLM-4.7, Gemma 4 26B A4B, DeepSeek V3.2 Exp (FP8), DeepSeek V3.2, Qwen3.6 35B A3B, gpt-oss-120b (high), Kimi K2 0905, GLM-4.6, GLM-4.7-Flash, Gemma 4 31B, DeepSeek V3.1 Terminus (FP8), DeepSeek V3.2 Exp (FP8), Qwen3 Coder Next (FP8), DeepSeek V3.1, DeepSeek V3.1, Qwen3 VL 235B A22B, Gemma 4 26B A4B, DeepSeek R1 0528, Qwen3 235B A22B 2507, Kimi K2, Qwen3 235B 2507, Qwen3 Coder 480B, MiniMax M1 80k, GLM-4.6V, GLM-4.7-Flash, DeepSeek V3 0324, gpt-oss-120b (low), gpt-oss-20b (high), gpt-oss-20b (low), Qwen3 VL 235B A22B, Llama 4 Maverick (FP8), NVIDIA Nemotron 3 Nano (FP4), DeepSeek V3 (Dec), DeepSeek V3 (Dec) Turbo, Ling 2.6 Flash, Qwen3 Next 80B A3B, GLM-4.6V, Qwen3 235B (FP8), Llama 4 Scout, Qwen3 VL 30B A3B, GLM-4.5V, Llama 3.1 8B, Gemma 3 27B, NVIDIA Nemotron 3 Nano (FP4), and GLM-4.5V.
Yes, provider performance can vary over time due to infrastructure changes, load balancing, and updates. We continuously benchmark all providers and display historical performance trends in the "Over Time" charts.
When choosing a model on Novita, consider: intelligence (for quality-sensitive tasks), output speed (for throughput-intensive tasks), latency (for interactive applications requiring quick first responses), pricing (for cost-sensitive workloads), and features like context window size, JSON mode, or function calling support.