Nebius: Models Intelligence, Performance & Price

Nebius
Nebius

Analysis of Nebius'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 Nebius for your use-case.

Most Intelligent

Updated
#1
GLM-5.2 (max) (FP4)
GLM-5.2 (max) (FP4)
51
#2
DeepSeek V4 Pro (max)
DeepSeek V4 Pro (max)
44
#3
Kimi K2.6
Kimi K2.6
44
#4
Kimi K2.7 Code (FP4)
Kimi K2.7 Code (FP4)
42
#5
DeepSeek V4 Pro (high)
DeepSeek V4 Pro (high)
41

Intelligence index

Total 30 models

Fastest

#1
NVIDIA Nemotron 3 Super
NVIDIA Nemotron 3 Super
367 t/s
#2
gpt-oss-120b (high) Base
gpt-oss-120b (high) Base
340 t/s
#3
gpt-oss-120b (low) Base
gpt-oss-120b (low) Base
317 t/s
#4
Nemotron 3 Nano Omni 30B A3B Reasoning (FP8)
Nemotron 3 Nano Omni 30B A3B Reasoning (FP8)
305 t/s
#5
Nemotron 3 Ultra
Nemotron 3 Ultra
284 t/s

Output speed

Total 30 models

Lowest Price

#1
Nemotron 3 Nano Omni 30B A3B Reasoning (FP8)
Nemotron 3 Nano Omni 30B A3B Reasoning (FP8)
$0.08
#2
NVIDIA Nemotron 3 Nano
NVIDIA Nemotron 3 Nano
$0.08
#3
Qwen3 32B Base
Qwen3 32B Base
$0.12
#4
Qwen3 30B A3B 2507
Qwen3 30B A3B 2507
$0.12
#5
Qwen3 32B Base
Qwen3 32B Base
$0.12

Blended price (per 1M tokens)

Total 30 models

Indicates a reasoning model

Nebius offers 30 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 Nebius are GLM-5.2 (max) (FP4) (51), DeepSeek V4 Pro (max) (44), Kimi K2.6 (44).
  • For output speed, the fastest models are NVIDIA Nemotron 3 Super (367 t/s), gpt-oss-120b (high) Base (340 t/s), gpt-oss-120b (low) Base (317 t/s).
  • For latency, GLM-5.2 (max) (FP4) (0.69s), gpt-oss-120b (high) Base (0.99s), Nemotron 3 Nano Omni 30B A3B Reasoning (FP8) (0.99s) offer the lowest time to first token.
  • For pricing, Nemotron 3 Nano Omni 30B A3B Reasoning (FP8) ($0.08), NVIDIA Nemotron 3 Nano ($0.08), Qwen3 32B Base ($0.12) offer the lowest blended prices per 1M tokens.
  • For context window size, DeepSeek V4 Pro (1m), DeepSeek V4 Pro (max) (1m), DeepSeek V4 Pro (high) (1m) support the largest context windows on Nebius.

Highlights

Updated

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 · Higher is better
Estimate (independent evaluation forthcoming)

Artificial Analysis Intelligence Index v4.1 includes: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

Figures represent performance of the model's first-party API (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

Intelligence Evaluations

Intelligence evaluations measured independently by Artificial Analysis · Higher is better
GDPval-AA v2

Agentic real-world work tasks, (Elo-500)/2000

𝜏³-Banking

Agentic tool use

Terminal-Bench v2.1

Agentic coding & terminal use

SciCode

Coding

Humanity's Last Exam

Reasoning & knowledge

GPQA Diamond

Scientific reasoning

CritPt

Physics reasoning

AA-Omniscience Accuracy

Knowledge

AA-Omniscience Non-Hallucination Rate

1 - hallucination rate

AA-LCR

Long context reasoning

AA-BriefcaseNew

Agentic knowledge work, Elo

AutomationBench-AANew

Agentic SaaS workflows

No data available
Harvey LAB-AANew

Legal agentic work, task all-pass rate

EnterpriseOps-Gym-AANew

Agentic business operations

IFBench

Instruction following

APEX-Agents-AA

Long-horizon agentic tasks

ITBench-AA

Kubernetes incident root-cause analysis

MMMU-Pro

Visual reasoning

Artificial Analysis Intelligence Index v4.1 includes: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

Intelligence vs. Price

Blended at 7:2:1 (cache-input-output) · Artificial Analysis Intelligence Index · USD per 1M tokens (blended)
Most attractive quadrant
Reasoning models are indicated by a lightbulb icon.

Artificial Analysis Intelligence Index v4.1 includes: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

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).

The blended cache price shown here uses cache hit price only. Other caching costs differ by provider:

  • Anthropic: charges a separate cache write fee, with different rates for 5-minute and 1-hour TTLs (1-hour TTL is more expensive).
  • Google (Vertex/Gemini): charges a per-hour cache storage fee in addition to cache hit pricing. Some providers also use tiered pricing for prompts above 200K tokens.
  • OpenAI, DeepSeek, others: typically charge only cache hit pricing with no write or storage fee.

See Prompt Caching for the full breakdown.

Figures represent performance of the model's first-party API (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

Context Window

Context Window

Context window: tokens limit · Higher is better

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

JSON Mode & Function Calling

Function (Tool) Calling & JSON Mode

ModelsFunction callingJSON Mode
GLM-5.2 (max) (FP4), Nebius logoGLM-5.2 (max) (FP4), Nebius
DeepSeek V4 Pro (max), Nebius logoDeepSeek V4 Pro (max), Nebius
Kimi K2.6, Nebius logoKimi K2.6, Nebius
Kimi K2.7 Code (FP4), Nebius logoKimi K2.7 Code (FP4), Nebius
DeepSeek V4 Pro (high), Nebius logoDeepSeek V4 Pro (high), Nebius
GLM-5.1 (FP8, Base), Nebius logoGLM-5.1 (FP8, Base), Nebius
Nemotron 3 Ultra, Nebius logoNemotron 3 Ultra, Nebius
GLM-5.1 (FP8, Base), Nebius logoGLM-5.1 (FP8, Base), Nebius
Kimi K2.6, Nebius logoKimi K2.6, Nebius
Qwen3.5 397B A17B (Base, FP4), Nebius logoQwen3.5 397B A17B (Base, FP4), Nebius
MiniMax-M2.5 (FP4), Nebius logoMiniMax-M2.5 (FP4), Nebius
Qwen3.5 397B A17B (Base, FP4), Nebius logoQwen3.5 397B A17B (Base, FP4), Nebius

Indicates whether the provider supports function calling in their API. Function calling is also known as 'Tool Calling'.

Indicates whether the provider supports JSON mode in their API. When JSON mode is enabled, the models will always return a valid JSON object.

Pricing

Intelligence vs. Price

Blended at 7:2:1 (cache-input-output) · Artificial Analysis Intelligence Index · USD per 1M tokens (blended)
Most attractive quadrant
Reasoning models are indicated by a lightbulb icon.

Artificial Analysis Intelligence Index v4.1 includes: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

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).

The blended cache price shown here uses cache hit price only. Other caching costs differ by provider:

  • Anthropic: charges a separate cache write fee, with different rates for 5-minute and 1-hour TTLs (1-hour TTL is more expensive).
  • Google (Vertex/Gemini): charges a per-hour cache storage fee in addition to cache hit pricing. Some providers also use tiered pricing for prompts above 200K tokens.
  • OpenAI, DeepSeek, others: typically charge only cache hit pricing with no write or storage fee.

See Prompt Caching for the full breakdown.

Figures represent performance of the model's first-party API (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

Performance Summary

Output Speed vs. Price

Output speed: output tokens per second · USD per 1M tokens (blended) · 10,000 input tokens
Most attractive quadrant
Nebius
Nebius (Base, FP4)
Nebius (FP4)
Nebius (FP8, Base)

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 · 10,000 input tokens

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 (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

Latency

Measured by Time (seconds) to First Token

Time to First Answer Token

Seconds to first token received · Lower is better

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.

Seconds to receive a 500 token response. Key components:

  • Input time: Time to receive the first response token
  • Thinking time (only for reasoning models): Time reasoning models spend outputting tokens to reason prior to providing an answer. Amount of tokens based on the average reasoning tokens across a diverse set of 60 prompts (methodology details).
  • Answer time: Time to generate 500 output tokens, based on output speed

Figures represent performance of the model's first-party API (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

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
Nebius
Nebius (Base, FP4)
Nebius (FP4)
Nebius (FP8, Base)

Seconds to receive a 500 token response. Key components:

  • Input time: Time to receive the first response token
  • Thinking time (only for reasoning models): Time reasoning models spend outputting tokens to reason prior to providing an answer. Amount of tokens based on the average reasoning tokens across a diverse set of 60 prompts (methodology details).
  • Answer time: Time to generate 500 output tokens, based on output speed

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).

Figures represent performance of the model's first-party API (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

Further Analysis
Nebius logoNebius
GLM-5.2 (max) (FP4)
432k
Open
51
$1.70
186
0.69
14.12
10.74
Model
Provider
Nebius logoNebius
DeepSeek V4 Pro (max)
1M
Open
44
$1.93
37
1.73
135.06
119.65
Model
Provider
Nebius logoNebius
Kimi K2.6
262k
Open
44
$1.25
226
1.92
23.85
19.72
Model
Provider
Nebius logoNebius
Kimi K2.7 Code (FP4)
256k
Open
42
$1.25
106
1.94
27.62
20.97
Model
Provider
Nebius logoNebius
DeepSeek V4 Pro (high)
1M
Open
41*
$1.93
50
1.75
51.95
40.13
Model
Provider
Nebius logoNebius
GLM-5.1 (FP8, Base)
200k
Open
40
$1.70
32
1.82
136.58
119.06
Model
Provider
Nebius logoNebius
Nemotron 3 Ultra
256k
Open
38
$1.20
284
2.22
12.00
8.02
Model
Provider
Nebius logoNebius
GLM-5.1 (FP8, Base)
200k
Open
35*
$1.70
34
1.86
16.53
--
Model
Provider
Nebius logoNebius
Kimi K2.6
262k
Open
35*
$1.25
230
1.95
4.12
--
Model
Provider
Nebius logoNebius
Qwen3.5 397B A17B (Base, FP4)
262k
Open
34
$0.90
106
2.00
36.75
30.04
Model
Provider
Nebius logoNebius
MiniMax-M2.5 (FP4)
196k
Open
34*
$0.39
83
1.81
32.07
24.21
Model
Provider
Nebius logoNebius
Qwen3.5 397B A17B (Base, FP4)
262k
Open
32*
$0.90
105
2.00
6.78
--
Model
Provider
Nebius logoNebius
DeepSeek V4 Pro
1.05M
Open
31*
$1.93
46
1.84
12.80
--
Model
Provider
Nebius logoNebius
NVIDIA Nemotron 3 Super
256k
Open
25
$0.36
367
1.84
8.65
5.45
Model
Provider
Nebius logoNebius
gpt-oss-120b (high) Base
128k
Open
24
$0.20
340
0.99
8.35
5.88
Model
Provider
Nebius logoNebius
Qwen3 235B 2507
262k
Open
18*
$0.24
79
1.33
7.63
--
Model
Provider
Nebius logoNebius
gpt-oss-120b (low) Base
128k
Open
18*
$0.20
317
1.02
8.90
6.31
Model
Provider
Nebius logoNebius
Qwen3 Next 80B A3B (FP8)
262k
Open
17
$0.26
101
1.03
25.71
19.74
Model
Provider
Nebius logoNebius
Nemotron 3 Nano Omni 30B A3B Reasoning (FP8)
65.5k
Open
15*
$0.08
305
0.99
9.19
6.55
Model
Provider
Nebius logoNebius
NVIDIA Nemotron 3 Nano
262k
Open
14
$0.08
132
1.06
20.04
15.19
Model
Provider
Nebius logoNebius
Qwen3 32B Base
32.8k
Open
12
$0.12
26
1.86
99.55
78.15
Model
Provider
Nebius logoNebius
Hermes 4 70B (FP8)
128k
Open
10*
$0.16
92
1.35
28.41
21.65
Model
Provider
Nebius logoNebius
Llama 3.3 70B Base
128k
Open
9
$0.16
24
2.40
23.56
--
Model
Provider
Nebius logoNebius
Llama Nemotron Ultra Base
131k
Open
9*
$0.72
51
2.39
51.16
39.02
Model
Provider
Nebius logoNebius
Qwen3 30B A3B 2507
262k
Open
9*
$0.12
69
1.04
8.31
--
Model
Provider
Nebius logoNebius
Hermes 4 405B (FP8)
128k
Open
9*
$1.20
43
2.38
60.60
46.57
Model
Provider
Nebius logoNebius
Hermes 4 405B (FP8)
128k
Open
9*
$1.20
41
2.38
14.56
--
Model
Provider
Nebius logoNebius
Qwen3 32B Base
32.8k
Open
9*
$0.12
27
1.94
20.49
--
Model
Provider
Nebius logoNebius
Gemma 3 27B (FP8)
110k
Open
7
$0.12
18
3.33
30.72
--
Model
Provider
Nebius logoNebius
Hermes 4 70B (FP8)
128k
Open
7*
$0.16
102
1.35
6.23
--
Model
Provider

Key definitions

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

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).

Time to first token received, in seconds, after API request sent. For reasoning models which share reasoning tokens, this will be the first reasoning token. For models which do not support streaming, this represents time to receive the completion.

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).

Price per token generated by the model (received from the API), represented as USD per million Tokens.

Price per token included in the request/message sent to the API, represented as USD per million Tokens.

Metrics are 'live' and are based on the past 72 hours of measurements, measurements are taken 8 times a day for single requests and 2 times per day for parallel requests.

Frequently Asked Questions

Common questions about Nebius

The most intelligent model available on Nebius is GLM-5.2 (max) with an Intelligence Index score of 51.

The fastest model on Nebius by output speed is NVIDIA Nemotron 3 Super at 367.1 tokens per second.

The model with the lowest time to first token on Nebius is GLM-5.2 (max) at 0.69s. Lower latency means faster initial response time.

The most affordable model on Nebius by blended price is Nemotron 3 Nano Omni 30B A3B Reasoning at $0.08 per 1M tokens (7:2:1 cache hit/input/output ratio).

Prices on Nebius vary up to 25x across models, from $0.08 per 1M tokens for Nemotron 3 Nano Omni 30B A3B Reasoning to $1.93 per 1M tokens for DeepSeek V4 Pro.

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

Yes, all 30 models on Nebius support JSON mode for structured output.

Yes, all 30 models on Nebius support function calling (tool use).

Yes, Nebius offers 19 reasoning models: GLM-5.2 (max), DeepSeek V4 Pro (max), Kimi K2.6, Kimi K2.7 Code, DeepSeek V4 Pro (high), GLM-5.1, Nemotron 3 Ultra, Qwen3.5 397B A17B, MiniMax-M2.5, NVIDIA Nemotron 3 Super, gpt-oss-120b (high), gpt-oss-120b (low), Qwen3 Next 80B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, NVIDIA Nemotron 3 Nano, Qwen3 32B, Hermes 4 70B, Llama Nemotron Ultra, and Hermes 4 405B. Reasoning models use extended thinking to work through complex problems before providing an answer.

Yes, all 30 models on Nebius 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 Nebius, 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.