The Known Good Updated 4 Sep 2026 Subscribe
Providers· Inference host· 80 tracked models

DeepInfra — pricing & performance

Compare models All providers Data sources
Models served
80
of 1,094 tracked models
Mean output speed
44
tokens per second, averaged over
79 measured endpoints
Mean latency
0.9s
time to first token, averaged over
79 measured endpoints
Cheapest blended
$0.022
per 1M tokens at a 3:1
input:output blend
Output speed on DeepInfra

Output tokens per second for the 79 of 80 models on DeepInfra with published throughput telemetry, the most recent figure OpenRouter publishes for each host. The full table below lists all 80, measured or not. Higher is better. We do not run these measurements ourselves.

⇩
200
152
135
130
123
111
98
97
91
87
87
79
The Known Good
80 models on DeepInfra
Full leaderboard →
Model Creator Input $/M Output $/M Blended Speed TTFT Context
DeepSeek: DeepSeek V3 DeepSeek $0.32 $0.89 $0.46 11 2.8s 164k
DeepSeek: DeepSeek V3 0324 DeepSeek $0.24 $0.90 $0.41 6 5.1s 164k
DeepSeek: DeepSeek V3.1 DeepSeek $0.25 $0.95 $0.42 16 0.7s 164k
DeepSeek: DeepSeek V3.1 Terminus DeepSeek $0.27 $0.95 $0.44 42 0.6s 164k
DeepSeek: DeepSeek V3.2 DeepSeek $0.26 $0.38 $0.29 13 1.6s 164k
DeepSeek: DeepSeek V4 Flash 0423 DeepSeek $0.090 $0.18 $0.11 44 0.6s 1.05M
DeepSeek: DeepSeek V4 Flash 0731 DeepSeek $0.080 $0.18 $0.10 32 0.7s 1.05M
DeepSeek: DeepSeek V4 Flash Vision Exp DeepSeek $0.22 $0.65 $0.32 49 0.9s 1.05M
DeepSeek: DeepSeek V4 Pro 0423 DeepSeek $1.30 $2.60 $1.62 43 1.0s 1.05M
DeepSeek: DeepSeek V4 Pro 0813 DeepSeek $1.30 $2.60 $1.62 58 0.9s 1.05M
DeepSeek: R1 0528 DeepSeek $0.50 $2.15 $0.91 18 0.9s 164k
Google: Gemma 3 12B Google $0.050 $0.15 $0.075 36 0.3s 131k
Google: Gemma 3 27B Google $0.080 $0.16 $0.10 16 0.7s 131k
Google: Gemma 3 4B Google $0.050 $0.10 $0.062 14 0.8s 131k
Google: Gemma 4 26B A4B Google $0.070 $0.34 $0.14 12 0.5s 262k
Google: Gemma 4 31B Google $0.090 $0.34 $0.15 72 2.3s 131k
IBM: Granite 4.2 8B IBM $0.060 $0.25 $0.11 39 0.2s 131k
Ling 3.0 Flash Fin inclusionai $0.060 $0.18 $0.090 54 0.4s 262k
Ling-3.0-flash inclusionai $0.060 $0.18 $0.090 33 2.7s 131k
Meta: Llama 3.1 70B Instruct Meta $0.40 $0.40 $0.40 18 0.3s 131k
Meta: Llama 3.1 8B Instruct Meta $0.020 $0.040 $0.025 22 0.5s 131k
Meta: Llama 3.3 70B Instruct Meta $0.10 $0.32 $0.15 14 0.6s 131k
Meta: Llama 4 Maverick Meta $0.20 $0.80 $0.35 38 0.4s 1.05M
Meta: Llama 4 Scout Meta $0.10 $0.30 $0.15 28 0.4s 328k
Meta: Llama Guard 4 12B Meta $0.18 $0.18 $0.18 9 0.3s 164k
Meta: Muse Glimmer 30B Meta $0.30 $1.20 $0.53 123 0.5s 131k
Microsoft: Phi 4 Microsoft $0.070 $0.14 $0.087 45 0.3s 16k
MiniMax: MiniMax M2.7 MiniMax $0.25 $1.00 $0.44 68 2.1s 197k
MiniMax: MiniMax M3 MiniMax $0.28 $1.10 $0.48 15 1.3s 524k
Mistral: Mistral Nemo Mistral $0.019 $0.030 $0.022 29 0.4s 131k
Mistral: Mistral Small 3 Mistral $0.050 $0.080 $0.058 43 0.3s 33k
Mistral: Mistral Small 3.2 24B Mistral $0.075 $0.20 $0.11 21 0.4s 128k
MoonshotAI: Kimi K2.5 Moonshot AI $0.45 $2.25 $0.90 43 0.8s 262k
MoonshotAI: Kimi K2.6 Moonshot AI $0.75 $3.50 $1.44 7 1.8s 262k
MoonshotAI: Kimi K2.7 Code Moonshot AI $0.68 $3.40 $1.36 55 0.6s 262k
MoonshotAI: Kimi K3 Moonshot AI $2.85 $14.25 $5.70 23 1.9s 1.05M
MythoMax 13B gryphe $0.40 $0.40 $0.40 45 0.3s 4k
Nous: Hermes 3 405B Instruct Nous Research $1.00 $1.00 $1.00 22 0.2s 131k
Nous: Hermes 3 70B Instruct Nous Research $0.70 $0.70 $0.70 31 0.4s 131k
NVIDIA: Nemotron 3 Nano 30B A3B NVIDIA $0.050 $0.20 $0.087 64 1.6s 262k
NVIDIA: Nemotron 3 Super NVIDIA $0.085 $0.40 $0.16 58 1.2s 262k
NVIDIA: Nemotron 3 Ultra NVIDIA $0.50 $2.20 $0.93 — — 262k
NVIDIA: Nemotron 3.5 Content Safety NVIDIA $0.20 $0.20 $0.20 111 0.2s 131k
NVIDIA: Nemotron 3.5 Lightning NVIDIA $0.080 $0.20 $0.11 200 0.3s 262k
OpenAI: gpt-oss-120b OpenAI $0.037 $0.17 $0.070 130 0.3s 131k
OpenAI: gpt-oss-20b OpenAI $0.030 $0.14 $0.058 87 0.3s 131k
Qwen2.5 72B Instruct Alibaba $0.36 $0.40 $0.37 17 0.5s 33k
Qwen: Qwen3 14B Alibaba $0.12 $0.24 $0.15 36 0.5s 41k
Qwen: Qwen3 235B A22B Instruct 2507 Alibaba $0.090 $0.55 $0.20 17 0.4s 262k
Qwen: Qwen3 235B A22B Thinking 2507 Alibaba $0.23 $2.30 $0.75 49 0.5s 262k
Qwen: Qwen3 30B A3B Alibaba $0.12 $0.50 $0.21 97 0.2s 41k
Qwen: Qwen3 32B Alibaba $0.080 $0.28 $0.13 21 0.5s 41k
Qwen: Qwen3 Coder 480B A35B Alibaba $0.30 $1.00 $0.47 48 1.0s 262k
Qwen: Qwen3 Next 80B A3B Instruct Alibaba $0.090 $1.10 $0.34 40 0.6s 262k
Qwen: Qwen3 VL 235B A22B Instruct Alibaba $0.20 $0.88 $0.37 9 0.6s 262k
Qwen: Qwen3 VL 30B A3B Instruct Alibaba $0.15 $0.60 $0.26 10 0.4s 262k
Qwen: Qwen3.5 397B A17B Alibaba $0.45 $3.00 $1.09 34 0.9s 262k
Qwen: Qwen3.5-122B-A10B Alibaba $0.29 $2.40 $0.82 37 0.9s 262k
Qwen: Qwen3.5-27B Alibaba $0.26 $2.60 $0.84 31 0.6s 262k
Qwen: Qwen3.5-35B-A3B Alibaba $0.14 $1.00 $0.35 98 0.4s 262k
Qwen: Qwen3.5-9B Alibaba $0.10 $0.15 $0.11 11 0.5s 262k
Qwen: Qwen3.6 27B Alibaba $0.32 $3.20 $1.04 53 0.4s 262k
Qwen: Qwen3.6 35B A3B Alibaba $0.10 $0.95 $0.31 27 0.8s 262k
Qwen: Qwen3.8 2.4T A95B Alibaba $2.00 $6.00 $3.00 87 0.8s 262k
Sao10K: Llama 3 8B Lunaris sao10k $0.040 $0.050 $0.043 79 0.2s 8k
Sao10K: Llama 3.1 Euryale 70B v2.2 sao10k $0.85 $0.85 $0.85 17 0.3s 131k
StepFun: Step 3.7 Flash stepfun $0.16 $0.92 $0.35 91 0.1s 262k
Tencent: Hy3 Tencent $0.14 $0.58 $0.25 67 0.3s 262k
Thinking Machines: Inkling Thinking Machines $0.95 $4.05 $1.73 135 0.4s 524k
Thinking Machines: Inkling Small Thinking Machines $0.45 $1.20 $0.64 152 4.5s 524k
Xiaomi: MiMo-V2.5 Xiaomi $0.13 $0.65 $0.26 3 0.7s 262k
Xiaomi: MiMo-V2.5-Pro Xiaomi $0.39 $1.17 $0.58 15 8.5s 1.05M
Z.ai: GLM 4.6 Z.ai $0.50 $2.00 $0.88 42 0.5s 203k
Z.ai: GLM 4.7 Z.ai $0.40 $1.75 $0.74 39 0.8s 203k
Z.ai: GLM 4.7 Flash Z.ai $0.060 $0.40 $0.14 34 0.6s 203k
Z.ai: GLM 5 Z.ai $0.60 $2.08 $0.97 22 0.8s 203k
Z.ai: GLM 5.1 Z.ai $1.05 $3.50 $1.66 31 0.9s 203k
Z.ai: GLM 5.2 Z.ai $0.49 $1.56 $0.76 37 1.4s 1.05M
Z.ai: GLM 5.3 Z.ai $1.20 $4.00 $1.90 28 1.6s 1.05M
Z.ai: GLM 5.3 Flash Z.ai $0.075 $0.25 $0.12 30 1.0s 1.05M
The Known Good
Source. Prices, endpoint availability and throughput for DeepInfra 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.