Novita: Models Intelligence, Performance & Price

Name: Intelligence
Creator: Artificial Analysis
License: https://artificialanalysis.ai/docs/legal/Terms-of-Use.pdf

Novita

This analysis is intended to support you in choosing the best model provided by Novita for your use-case.

Most Intelligent

DeepSeek V4.1 Flash 1017 (max)

GLM-5.3 (max)

GLM-5.3-Flash

DeepSeek V4.1 Flash (max)

DeepSeek V4 Pro 0813 (max)

Intelligence Index

Total 60 models

Fastest

gpt-oss-120b (low)

240 t/s

gpt-oss-120b (high)

236 t/s

Nemotron 3 Nano (FP4)

236 t/s

Nemotron 3 Nano (non-reasoning) (FP4)

222 t/s

DeepSeek V4.1 Flash (max)

221 t/s

Output speed

Total 60 models

Lowest Price

Llama 3.1 8B

$0.02

GLM-4.7-Flash

$0.06

GLM-4.7-Flash (non-reasoning)

$0.06

Nemotron 3 Nano (FP4)

$0.07

Nemotron 3 Nano (non-reasoning) (FP4)

$0.07

Blended price (per 1M tokens)

Total 60 models

Novita offers 60 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 DeepSeek V4.1 Flash 1017 (max) (45), GLM-5.3 (max) (45), and GLM-5.3-Flash (42).
For output speed, the fastest models are gpt-oss-120b (low) (240 t/s), gpt-oss-120b (high) (236 t/s), and Nemotron 3 Nano (FP4) (236 t/s).
For latency, Llama 3.1 8B (1.01s), Nemotron 3 Nano (non-reasoning) (FP4) (1.04s), and Gemma 4 26B A4B (non-reasoning) (1.05s) offer the lowest time to first answer token.
For pricing, Llama 3.1 8B ($0.02), GLM-4.7-Flash ($0.06), and GLM-4.7-Flash (non-reasoning) ($0.06) offer the lowest blended prices per 1M tokens. Prices vary up to 2.8x across models.
For context window size, DeepSeek V4.1 Flash 1017 (max) (1M), GLM-5.3-Flash (1M), and DeepSeek V4 Pro 0813 (max) (1M) support the largest context windows on Novita.

Intelligence

Artificial Analysis Intelligence Index · Higher is better

Speed

Output tokens per second · Higher is better

Price

USD per 1M tokens (blended) · Lower is better

Intelligence Evaluations

Artificial Analysis Intelligence Index

Artificial Analysis Intelligence Index v4.3.2 incorporates 10 evaluations: AA-Briefcase v1.1, GDPval-AA v2.1, AutomationBench-AA, Terminal-Bench 4.0, SciCode, Humanity's Last Exam, GDP.pdf, CritPt, AA-Omniscience, AA-LCR v1.1

Estimate (independent evaluation forthcoming)

Artificial Analysis Intelligence Index v4.3.2 includes: AA-Briefcase v1.1, GDPval-AA v2.1, AutomationBench-AA, Terminal-Bench 4.0, SciCode, Humanity's Last Exam, GDP.pdf, CritPt, AA-Omniscience, AA-LCR v1.1. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

Evaluations

Evaluations measured independently by Artificial Analysis · AA-PersonalityBench scores are percentile ranks from 1 to 100

AA-Briefcase v1.1

Agentic knowledge work, Elo

AA-Omniscience Accuracy

Knowledge

AA-Omniscience Non-Hallucination Rate

1 - hallucination rate

AutomationBench-AA

Agentic SaaS workflows

GDPval-AA v2.1

Agentic real-world work tasks, Elo

Humanity's Last Exam

Reasoning & knowledge

GDP.pdf

Professional document reasoning, All-pass

Terminal-Bench 4.0

Agentic coding & terminal use

AA-LCR v1.1

Long context reasoning

SciCodeUnder review

Coding

CritPt

Physics reasoning

Terminal-Bench-Science 0.1New

Agentic scientific research workflows in a terminal

FrontierCode 1.1New

Agentic software engineering in open-source repositories

AA-AnalystAgent

Quantitative analysis on spreadsheets & documents

Harvey LAB-AA v1.1Updated

Legal agentic work, Hallucination-Gated All-Pass Rate

EnterpriseOps-Gym-AA

Agentic business operations

AA-3DNew

Agentic interactive 3D experiences, Elo

ITBench-AA

Kubernetes incident root-cause analysis

MLCR-AAUpdated

Medical long context reasoning

MMMU-Pro

Visual reasoning

AA-PersonalityBench: Expressiveness ScoreNew

Rich, varied, well-crafted language

AA-PersonalityBench: Organization ScoreNew

Well-structured, scannable, answer-first

AA-PersonalityBench: Transparency ScoreNew

Truthful, acknowledges uncertainty, non-sycophantic

AA-PersonalityBench: Slop ScoreNew

Use of filler and generic language

While model intelligence generally translates across use cases, specific evaluations may be more relevant for certain use cases.

Intelligence Index vs. Price

Blended at 7:2:1 (cache-input-output) · USD per 1M tokens (blended)

Most attractive quadrant

Pareto line

While higher intelligence models are typically more expensive, they do not all follow the same price-quality curve.

Context Window

Context window: tokens limit · Higher is better

Larger context windows are relevant to RAG (Retrieval Augmented Generation) LLM workflows which typically involve reasoning and information retrieval of large amounts of data.

Maximum number of combined input & output tokens. Output tokens commonly have a significantly lower limit (varied by model).

Pricing

Intelligence Index vs. Price

Blended at 7:2:1 (cache-input-output) · USD per 1M tokens (blended)

Most attractive quadrant

Pareto line

While higher intelligence models are typically more expensive, they do not all follow the same price-quality curve.

Performance Summary

Output Speed vs. Price

Output speed: output tokens per second · USD per 1M tokens (blended)

Most attractive quadrant

Pareto line

Tokens per second received while the model is generating tokens (ie. after first chunk has been received from the API for models which support streaming).

Price per token, shown in USD per million tokens. Price is a blend of cache hit, input, and output token prices using the selected ratio (default 7:2:1 cache-input-output).

Speed

Measured by Output Speed (tokens per second)

Output Speed

Output tokens per second · Higher is better

Tokens per second received while the model is generating tokens (ie. after first chunk has been received from the API for models which support streaming).

Figures represent performance of the model's first-party API or the median across providers where a first-party API is not available.

Latency

Measured by Time (seconds) to First Token

Latency: Time To First Answer Token

Seconds to first answer token received · Accounts for reasoning model 'thinking' time

Time to first answer token received, in seconds, after API request sent. For reasoning models, this includes the 'thinking' time of the model before providing an answer. For models which do not support streaming, this represents time to receive the completion.

Cache Behaviour

Cache Hit Rate

Share of cacheable input tokens served from cache · Median of the last four weeks, updated Sep 26, 2026

The share of cacheable input tokens served from the provider's prompt cache, measured during multi-turn agentic coding sessions. We regularly measure each endpoint and report a representative rate based on recent measurements. Higher values mean more of the input is billed at the cheaper cache-hit price, where the provider offers one, which lowers the effective cost of agentic workloads. For further details, see our methodology page.

Cost per Task vs. Cache Hit Rate

Weighted average cost (USD) per Intelligence Index task · Cache hit rate: median of the last four weeks, updated Sep 26, 2026

Most attractive quadrant

Pareto line

Weighted average cost per Intelligence Index task. Each evaluation’s cost is calculated from input, cache hit, cache write, reasoning, and answer token prices, divided by task count, and weighted by its Intelligence Index weight.

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

End-to-end response time: end-to-end seconds to output 500 tokens · USD per 1M tokens (blended)

Most attractive quadrant

Pareto line

Price per token, shown in USD per million tokens. Price is a blend of cache hit, input, and output token prices using the selected ratio (default 7:2:1 cache-input-output).


									Further Analysis
MiMo-V2.6-Pro	1M	Open	46	$0.13	44	2.38	58.66	45.02	Model Provider
DeepSeek V4.1 Flash 1017 (max)	1.05M	Open	45	$0.24	166	1.64	16.72	12.06	Model Provider
GLM-5.3 (max)	1M	Open	45	$2.00	85	2.19	31.61	23.54	Model Provider
GLM-5.3-Flash	1.05M	Open	42	$0.29	73	2.92	37.24	27.46	Model Provider
DeepSeek V4.1 Flash (max)	524k	Open	39	$0.24	233	1.42	12.15	8.58	Model Provider
MiMo-V2.6-Flash	1M	Open	38	$0.06	41	4.34	65.58	48.99	Model Provider
DeepSeek V4 Pro 0813 (max)	1.05M	Open	36	$0.70	64	2.26	41.45	31.35	Model Provider
DeepSeek V4 Flash 0731 (max)	1.05M	Open	34	$0.26	139	1.67	19.59	14.34	Model Provider
GLM-5.2 (max) (FP8)	1.05M	Open	34	$0.88	69	2.67	39.13	29.17	Model Provider
DeepSeek V4 Pro (max)	1.05M	Open	30	$1.09	91	1.99	55.50	48.02	Model Provider
DeepSeek V4 Pro (high)	1.05M	Open	30*	--	--	--	--	--	Model Provider
MiniMax-M3	1M	Open	29	$0.46	91	1.99	29.37	21.90	Model Provider
GLM-5 FP8	203k	Open	28*	--	63	2.34	59.81	49.50	Model Provider
Kimi K2.6	262k	Open	27	$0.67	48	3.98	108.18	93.67	Model Provider
GLM-5.1 (FP8)	205k	Open	26	$1.07	56	2.65	79.45	67.85	Model Provider
MiMo-V2.5-Pro	1.05M	Open	26	$0.07	45	2.07	57.90	44.66	Model Provider
Kimi K2.7 Code	262k	Open	26	$0.55	52	3.80	56.46	43.01	Model Provider
Hy3	262k	Open	25	$0.07	72	2.49	37.18	27.75	Model Provider
GLM-5.1 (non-reasoning) (FP8)	205k	Open	24*	--	38	3.82	17.01	--	Model Provider
Kimi K2.6 (non-reasoning)	262k	Open	24*	--	46	4.43	15.20	--	Model Provider
Kimi K2.5	262k	Open	23*	--	49	3.99	74.68	60.49	Model Provider
Qwen3.5 27B	262k	Open	23*	--	54	6.30	52.20	36.72	Model Provider
MiniMax-M2.5	205k	Open	23*	--	58	2.03	45.37	34.67	Model Provider
MiniMax-M2.7 (FP8)	205k	Open	23	--	64	1.79	47.73	38.19	Model Provider
GLM-4.7	205k	Open	22*	--	34	5.00	77.62	58.09	Model Provider
Kimi K2 Thinking	262k	Open	22*	--	50	2.26	52.03	39.82	Model Provider
MiMo-V2.5	1.05M	Open	22	$0.02	38	3.35	69.11	52.60	Model Provider
GLM-5 (non-reasoning) (FP8)	203k	Open	22*	--	61	2.25	10.46	--	Model Provider
Qwen3.5 397B A17B (non-reasoning)	262k	Open	21*	--	57	2.25	11.08	--	Model Provider
Qwen3.6 27B	262k	Open	21	$0.62	56	3.39	113.56	101.24	Model Provider
Qwen3.6 27B (non-reasoning)	262k	Open	20*	--	59	3.46	11.89	--	Model Provider
Kimi K2.5 (non-reasoning)	262k	Open	19*	--	47	1.99	12.64	--	Model Provider
Qwen3.5 35B A3B	262k	Open	19*	--	79	2.13	33.90	25.42	Model Provider
GLM-4.6	205k	Open	19*	--	35	4.45	75.66	56.97	Model Provider
Qwen3.5 397B A17B	262k	Open	18	$0.47	56	2.22	68.62	57.39	Model Provider
MiMo-V2.5-Pro (non-reasoning)	1.05M	Open	18*	--	40	2.06	14.58	--	Model Provider
Qwen3.6 35B A3B	205k	Open	18	$0.31	133	1.96	46.34	40.62	Model Provider
GLM-4.7 (non-reasoning)	205k	Open	17*	--	34	4.32	19.24	--	Model Provider
Gemma 4 26B A4B	262k	Open	17*	--	153	1.45	17.77	13.05	Model Provider
Qwen3.5 122B A10B	262k	Open	16	$0.32	97	2.48	28.21	20.58	Model Provider
Kimi K2 0905	262k	Open	15*	--	46	1.40	12.33	--	Model Provider
Qwen3.6 35B A3B (non-reasoning)	205k	Open	15*	--	140	1.95	5.51	--	Model Provider
GLM-4.6 (non-reasoning)	205k	Open	15*	--	33	4.34	19.37	--	Model Provider
GLM-4.7-Flash	200k	Open	15*	--	78	3.26	35.46	25.75	Model Provider
DeepSeek V3.1 Terminus (FP8)	131k	Open	15	--	--	--	--	--	Model Provider
Gemma 4 31B	262k	Open	15	$0.10	25	2.11	91.68	69.54	Model Provider
DeepSeek V3.1 Terminus (non-reasoning) (FP8)	131k	Open	14*	--	--	--	--	--	Model Provider
Gemma 4 31B (non-reasoning)	262k	Open	14*	--	22	1.99	25.15	--	Model Provider
DeepSeek V3.1 (non-reasoning)	164k	Open	14*	--	--	--	--	--	Model Provider
DeepSeek V3.1	131k	Open	13*	--	--	--	--	--	Model Provider
Gemma 4 26B A4B (non-reasoning)	262k	Open	13*	--	98	1.13	6.23	--	Model Provider
DeepSeek R1 0528	164k	Open	13*	--	29	1.43	86.95	68.42	Model Provider
Kimi K2	131k	Open	13*	--	47	1.96	12.60	--	Model Provider
MiniMax M1 80k	1M	Open	12*	--	72	1.84	36.67	27.87	Model Provider
gpt-oss-120b (high)	131k	Open	12	$0.04	213	1.02	12.76	9.39	Model Provider
DeepSeek R1 (Jan) Turbo	64k	Open	11	$0.11	--	--	--	--	Model Provider
DeepSeek R1 (Jan)	64k	Open	11	$0.46	--	--	--	--	Model Provider
GLM-4.6V	131k	Open	11*	--	88	4.05	32.42	22.70	Model Provider
GLM-4.7-Flash (non-reasoning)	200k	Open	11*	--	64	3.82	11.61	--	Model Provider
gpt-oss-120b (low)	131k	Open	10*	--	238	0.95	11.46	8.41	Model Provider
Llama 4 Maverick (FP8)	1.05M	Open	9	$0.04	35	1.21	15.40	--	Model Provider
Nemotron 3 Nano (FP4)	262k	Open	9	$0.02	228	1.07	12.04	8.77	Model Provider
DeepSeek V3 (Dec)	64k	Open	8	$0.05	86	1.39	7.21	--	Model Provider
GLM-4.6V (non-reasoning)	131k	Open	8*	--	81	3.94	10.13	--	Model Provider
Qwen3 VL 30B A3B	131k	Open	8*	--	38	2.89	16.05	--	Model Provider
Llama 3.3 70B	131k	Open	8*	--	41	1.79	13.95	--	Model Provider
GLM-4.5V	65.5k	Open	8*	--	49	3.40	54.25	40.68	Model Provider
Llama 3.1 8B	16.4k	Open	7*	--	156	0.96	4.16	--	Model Provider
Nemotron 3 Nano (non-reasoning) (FP4)	262k	Open	7*	--	235	1.07	3.20	--	Model Provider
GLM-4.5V (non-reasoning)	65.5k	Open	7*	--	50	3.30	13.23	--	Model Provider
Llama 4 Scout	131k	Open	6	$0.11	18	1.39	28.53	--	Model Provider
Llama 3 70B	8.19k	Open	5*	--	--	--	--	--	Model Provider
Gemma 3 27B	98.3k	Open	5	$0.24	36	1.70	15.57	--	Model Provider
Llama 3 8B	8.19k	Open	5*	--	--	--	--	--	Model Provider

Key definitions

Maximum number of combined input & output tokens. Output tokens commonly have a significantly lower limit (varied by model).

Frequently Asked Questions

Common questions about Novita

Novita offers 60 models that we track: DeepSeek V4.1 Flash 1017 (max), GLM-5.3 (max), GLM-5.3-Flash, DeepSeek V4.1 Flash (max), DeepSeek V4 Pro 0813 (max), DeepSeek V4 Flash 0731 (max), GLM-5.2 (max) (FP8), DeepSeek V4 Pro (max), MiniMax-M3, GLM-5 FP8, Kimi K2.6, GLM-5.1 (FP8), Kimi K2.7 Code, Hy3, GLM-5.1 (non-reasoning) (FP8), Kimi K2.6 (non-reasoning), Kimi K2.5, Qwen3.5 27B, MiniMax-M2.5, MiniMax-M2.7 (FP8), GLM-4.7, Kimi K2 Thinking, GLM-5 (non-reasoning) (FP8), Qwen3.5 397B A17B (non-reasoning), Qwen3.6 27B, Qwen3.6 27B (non-reasoning), Kimi K2.5 (non-reasoning), Qwen3.5 35B A3B, GLM-4.6, Qwen3.5 397B A17B, Qwen3.6 35B A3B, GLM-4.7 (non-reasoning), Gemma 4 26B A4B, Qwen3.5 122B A10B, Kimi K2 0905, Qwen3.6 35B A3B (non-reasoning), GLM-4.6 (non-reasoning), GLM-4.7-Flash, Gemma 4 31B, Gemma 4 31B (non-reasoning), Gemma 4 26B A4B (non-reasoning), DeepSeek R1 0528, Kimi K2, MiniMax M1 80k, gpt-oss-120b (high), GLM-4.6V, GLM-4.7-Flash (non-reasoning), gpt-oss-120b (low), Llama 4 Maverick (FP8), Nemotron 3 Nano (FP4), DeepSeek V3 (Dec), GLM-4.6V (non-reasoning), Qwen3 VL 30B A3B, Llama 3.3 70B, GLM-4.5V, Llama 3.1 8B, Nemotron 3 Nano (non-reasoning) (FP4), GLM-4.5V (non-reasoning), Llama 4 Scout, and Gemma 3 27B.

The most intelligent model available on Novita is DeepSeek V4.1 Flash 1017 (max) with an Intelligence Index score of 45.

The fastest model on Novita by output speed is gpt-oss-120b (low) at 240.0 tokens per second.

The model with the lowest time to first answer token on Novita is Llama 3.1 8B at 1.01s. 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 39x across models, from $0.02 per 1M tokens for Llama 3.1 8B to $0.90 per 1M tokens for GLM-5.2 (max) (FP8).

Yes, Novita offers an OpenAI-compatible API, making it easy to switch from OpenAI or use existing OpenAI SDK integrations.

59 of 60 models on Novita support JSON mode for structured output.

56 of 60 models on Novita support function calling (tool use).

Yes, Novita offers 36 reasoning models: DeepSeek V4.1 Flash 1017 (max), GLM-5.3 (max), GLM-5.3-Flash, DeepSeek V4.1 Flash (max), DeepSeek V4 Pro 0813 (max), DeepSeek V4 Flash 0731 (max), GLM-5.2 (max) (FP8), DeepSeek V4 Pro (max), MiniMax-M3, GLM-5 FP8, Kimi K2.6, GLM-5.1 (FP8), Kimi K2.7 Code, Hy3, Kimi K2.5, Qwen3.5 27B, MiniMax-M2.5, MiniMax-M2.7 (FP8), GLM-4.7, Kimi K2 Thinking, Qwen3.6 27B, Qwen3.5 35B A3B, GLM-4.6, Qwen3.5 397B A17B, Qwen3.6 35B A3B, Gemma 4 26B A4B, Qwen3.5 122B A10B, GLM-4.7-Flash, Gemma 4 31B, DeepSeek R1 0528, MiniMax M1 80k, gpt-oss-120b (high), GLM-4.6V, gpt-oss-120b (low), Nemotron 3 Nano (FP4), and GLM-4.5V. Reasoning models use extended thinking to work through complex problems before providing an answer.

Yes, all 60 models on Novita are open weight models.

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.

Novita: Models Intelligence, Performance & Price

Most Intelligent

Fastest

Lowest Price

Intelligence