The Known Good Updated 9 Sep 2026 Subscribe
Providers· Inference host· 17 tracked models

Nebius — pricing & performance

Compare models All providers Data sources
Models served
17
of 1,111 tracked models
Mean output speed
52
tokens per second, averaged over
15 measured endpoints
Mean latency
2.3s
time to first token, averaged over
15 measured endpoints
Cheapest blended
$0.10
per 1M tokens at a 3:1
input:output blend
Output speed on Nebius

Output tokens per second for the 15 of 17 models on Nebius with published throughput telemetry, the most recent figure OpenRouter publishes for each host. The full table below lists all 17, measured or not. Higher is better. We do not run these measurements ourselves.

⇩
221
168
90
77
35
33
30
30
24
20
16
15
The Known Good
17 models on Nebius
Full leaderboard →
Model Creator Input $/M Output $/M Blended Speed TTFT Context
DeepSeek: DeepSeek V4 Flash 0731 DeepSeek $0.14 $0.28 $0.17 24 5.2s 1.05M
Google: Gemma 3 27B Google $0.10 $0.30 $0.15 9 1.5s 110k
Meta: Llama 3.3 70B Instruct Meta $0.13 $0.40 $0.20 5 4.7s 131k
MiniMax: MiniMax M2.5 MiniMax $0.30 $1.20 $0.53 77 0.8s 197k
MoonshotAI: Kimi K2.7 Code Moonshot AI $0.95 $4.00 $1.71 16 1.0s 262k
MoonshotAI: Kimi K3 Moonshot AI $3.00 $15.00 $6.00 7 16.0s 1.05M
Nous: Hermes 4 405B Nous Research $1.00 $3.00 $1.50 30 0.3s 131k
Nous: Hermes 4 70B Nous Research $0.13 $0.40 $0.20 — — 131k
NVIDIA: Nemotron 3 Nano 30B A3B NVIDIA $0.060 $0.24 $0.10 168 0.3s 262k
NVIDIA: Nemotron 3 Super NVIDIA $0.30 $0.90 $0.45 — — 8k
OpenAI: gpt-oss-120b OpenAI $0.15 $0.60 $0.26 221 0.3s 131k
Qwen: Qwen2.5 VL 72B Instruct Alibaba $0.25 $0.75 $0.38 15 1.1s 32k
Qwen: Qwen3 235B A22B Instruct 2507 Alibaba $0.20 $0.60 $0.30 35 0.6s 262k
Qwen: Qwen3 30B A3B Instruct 2507 Alibaba $0.10 $0.30 $0.15 33 0.3s 262k
Qwen: Qwen3 32B Alibaba $0.10 $0.30 $0.15 20 0.5s 41k
Qwen: Qwen3 Next 80B A3B Thinking Alibaba $0.15 $1.20 $0.41 90 0.4s 128k
Z.ai: GLM 5.1 Z.ai $1.40 $4.40 $2.15 30 1.1s 203k
The Known Good
Source. Prices, endpoint availability and throughput for Nebius are ingested from OpenRouter; model metadata is enriched from models.dev. A blank cell means the figure was not published for that endpoint — it never means zero. Blended price is (3 × input + output) ÷ 4, the same 3:1 blend used everywhere on this site; see methodology. Quality scores are not provider-specific: a model scores the same wherever it is hosted, so quality lives on the model pages.