Hugging Face's logo

Inference Providers

·

Metrics for top trending models

Browse all models Learn more
Kimi-K2-Thinking
novita
cheapest
0.6 2.5 262144 0.75 34 Yes No
Kimi-K2-Thinking
together
fastest
1.2 4 262144 1.79 59 Yes Yes
DeepSeek-OCR
novita
0.03 0.03 8192 0.68 164 No Yes
MiniMax-M2
novita
0.3 1.2 204800 0.98 60 No No
GLM-4.6
novita
cheapest
0.6 2.2 204800 0.67 41 Yes No
GLM-4.6
cerebras
fastest
- - - 0.31 323 Yes No
GLM-4.6
zai-org
- - - 3.79 40 Yes No
Llama-3.1-8B-Instruct
novita
cheapest
0.02 0.05 16384 0.73 86 No No
Llama-3.1-8B-Instruct
nebius
0.03 0.09 131072 0.43 174 No Yes
Llama-3.1-8B-Instruct
cerebras
fastest
0.1 0.1 - 0.33 1151 No No
Llama-3.1-8B-Instruct
sambanova
0.1 0.2 16384 0.75 503 Yes Yes
Llama-3.1-8B-Instruct
nscale
0.06 0.06 131072 0.88 59 No Yes
Llama-3.1-8B-Instruct
fireworks-ai
0.2 0.2 131072 0.41 216 No No
Llama-3.1-8B-Instruct
scaleway
- - - 0.57 98 Yes Yes
gpt-oss-20b
groq
fastest
0.1 0.5 131072 0.17 851 Yes No
gpt-oss-20b
novita
0.04 0.15 131072 0.91 160 No Yes
gpt-oss-20b
nebius
0.05 0.2 131072 0.37 202 Yes Yes
gpt-oss-20b
nscale
0.05 0.2 131072 0.85 123 Yes Yes
gpt-oss-20b
hyperbolic
cheapest
0.1 0.1 131072 0.55 143 No No
gpt-oss-20b
together
0.05 0.2 131072 0.28 164 Yes Yes
gpt-oss-20b
fireworks-ai
0.05 0.2 131072 1.01 148 Yes No
Kimi-K2-Instruct-0905
groq
fastest
- - 262144 0.26 165 Yes No
Kimi-K2-Instruct-0905
novita
cheapest
0.6 2.5 262144 0.78 37 Yes No
Kimi-K2-Instruct-0905
together
1 3 262144 1.46 50 Yes Yes
gpt-oss-120b
groq
0.15 0.75 131072 0.24 417 Yes No
gpt-oss-120b
novita
cheapest
0.05 0.25 131072 0.90 139 No Yes
gpt-oss-120b
nebius
0.15 0.6 131072 0.38 164 Yes Yes
gpt-oss-120b
cerebras
fastest
0.25 0.69 - 0.28 1035 Yes No
gpt-oss-120b
sambanova
0.22 0.59 131072 1.18 377 Yes Yes
gpt-oss-120b
nscale
0.1 0.4 131072 0.98 94 Yes Yes
gpt-oss-120b
hyperbolic
0.3 0.3 131072 0.74 268 Yes No
gpt-oss-120b
together
0.15 0.6 131072 0.37 138 Yes Yes
gpt-oss-120b
fireworks-ai
0.15 0.6 131072 0.52 100 No No
gpt-oss-120b
scaleway
- - - 0.49 151 Yes Yes
Kimi-K2-Instruct
novita
cheapest
0.57 2.3 131072 1.24 26 Yes Yes
Kimi-K2-Instruct
nebius
fastest
0.5 2.4 131072 0.41 60 Yes Yes
Kimi-K2-Instruct
together
1 3 131072 1.19 49 Yes Yes
Kimi-K2-Instruct
fireworks-ai
0.6 2.5 131072 1.03 49 Yes No
Qwen3-VL-8B-Instruct
novita
0.08 0.5 131072 0.73 56 Yes Yes
Qwen3-4B-Instruct-2507
nscale
0.01 0.03 262144 0.97 86 Yes No
Qwen3-8B
nscale
0.07 0.18 40960 0.92 59 Yes No
Llama-3.2-3B-Instruct
novita
cheapest
0.03 0.05 32768 0.78 122 Yes No
Llama-3.2-3B-Instruct
sambanova
- - - - - - -
Llama-3.2-3B-Instruct
hyperbolic
0.1 0.1 131072 1.09 128 No No
Llama-3.2-3B-Instruct
together
fastest
0.06 0.06 131072 0.28 151 Yes Yes
Qwen3-VL-30B-A3B-Instruct
novita
0.2 0.7 131072 1.62 81 Yes Yes
Meta-Llama-3-8B-Instruct
groq
- - - - - - -
Meta-Llama-3-8B-Instruct
novita
0.04 0.04 8192 0.74 85 No No
Meta-Llama-3-8B-Instruct
together
- - - - - - -
DeepSeek-R1
novita
0.7 2.5 64000 0.96 27 Yes No
DeepSeek-R1
sambanova
fastest
- - - 2.82 139 Yes Yes
DeepSeek-R1
hyperbolic
cheapest
2 2 163840 1.32 43 No No
DeepSeek-R1
together
3 7 163840 2.75 38 No Yes
DeepSeek-R1
fireworks-ai
- - - - - - -
SmolLM3-3B
hf-inference
- - - 0.22 87 Yes No
Qwen3-Coder-30B-A3B-Instruct
nebius
cheapest fastest
0.1 0.3 262144 0.44 121 Yes Yes
Qwen3-Coder-30B-A3B-Instruct
fireworks-ai
- - - - - - -
Qwen3-Coder-30B-A3B-Instruct
scaleway
- - - 0.38 108 Yes No
Qwen2.5-7B-Instruct
together
0.3 0.3 32768 0.26 138 Yes Yes
Llama-3.3-70B-Instruct
groq
0.59 0.79 131072 0.23 300 Yes No
Llama-3.3-70B-Instruct
novita
cheapest
0.13 0.39 131072 0.53 32 Yes No
Llama-3.3-70B-Instruct
nebius
0.25 0.75 131072 0.52 132 Yes Yes
Llama-3.3-70B-Instruct
cerebras
fastest
0.85 1.2 - 0.31 1028 Yes No
Llama-3.3-70B-Instruct
sambanova
0.6 1.2 131072 0.43 292 Yes Yes
Llama-3.3-70B-Instruct
nscale
0.4 0.4 131072 0.97 16 No Yes
Llama-3.3-70B-Instruct
hyperbolic
0.4 0.4 131072 1.14 98 No No
Llama-3.3-70B-Instruct
together
0.88 0.88 131072 0.44 110 Yes Yes
Llama-3.3-70B-Instruct
fireworks-ai
0.9 0.9 131072 0.81 142 No No
Llama-3.3-70B-Instruct
scaleway
- - - 0.67 37 Yes Yes
Apertus-8B-Instruct-2509
publicai
- - - 0.96 109 No Yes
DeepSeek-V3.2-Exp
novita
0.27 0.41 163840 0.91 30 Yes Yes
Llama-3.2-1B-Instruct
novita
- - 131000 0.82 201 No No
Llama-3.2-1B-Instruct
sambanova
- - - - - - -
Qwen3-14B
nscale
0.07 0.2 40960 1.02 42 Yes No
Qwen3-30B-A3B-Instruct-2507
nebius
0.1 0.3 262144 0.45 103 Yes Yes
Qwen2.5-VL-7B-Instruct
hyperbolic
0.2 0.2 32768 0.53 109 No No
GLM-4.5-Air
novita
cheapest
0.13 0.85 131072 0.80 65 Yes No
GLM-4.5-Air
nebius
fastest
0.2 1.2 131072 0.41 69 Yes Yes
GLM-4.5-Air
fireworks-ai
- - - - - - -
GLM-4.5-Air
zai-org
- - - 3.96 55 Yes No
gemma-3-27b-it
nebius
cheapest fastest
0.2 0.6 110000 0.46 77 No Yes
gemma-3-27b-it
scaleway
- - - 0.64 43 Yes No
gpt-oss-safeguard-20b
groq
- - 131072 0.53 605 Yes No
DeepSeek-R1-Distill-Qwen-1.5B
nscale
0.1 0.1 131072 1.06 136 No No
GLM-4.5V
novita
cheapest fastest
0.6 1.8 65536 2.14 49 Yes No
GLM-4.5V
zai-org
- - - 3.34 46 Yes No
Qwen3-235B-A22B-Instruct-2507
novita
cheapest
0.09 0.58 131072 1.52 25 Yes Yes
Qwen3-235B-A22B-Instruct-2507
nebius
0.2 0.6 262144 1.81 21 Yes Yes
Qwen3-235B-A22B-Instruct-2507
cerebras
fastest
0.6 1.2 - 0.29 499 Yes No
Qwen3-235B-A22B-Instruct-2507
nscale
0.2 0.6 32768 1.13 23 Yes Yes
Qwen3-235B-A22B-Instruct-2507
hyperbolic
2 2 262144 1.12 53 Yes No
Qwen3-235B-A22B-Instruct-2507
together
0.2 0.6 262144 0.44 46 Yes Yes
Qwen3-235B-A22B-Instruct-2507
fireworks-ai
0.22 0.88 262144 2.19 24 Yes No
Qwen3-235B-A22B-Instruct-2507
scaleway
- - - 0.53 72 Yes Yes
DeepSeek-R1-0528-Qwen3-8B
novita
0.06 0.09 128000 0.77 71 No No
DeepSeek-V3
novita
0.4 1.3 64000 1.30 29 Yes No
DeepSeek-V3
together
cheapest fastest
1.25 1.25 131072 1.14 36 Yes Yes
DeepSeek-V3
fireworks-ai
- - - - - - -
GLM-4.5
novita
0.6 2.2 131072 0.63 44 Yes No
GLM-4.5
nebius
0.6 2.2 131072 0.40 36 Yes No
GLM-4.5
fireworks-ai
cheapest fastest
0.55 2.19 131072 0.85 49 Yes No
GLM-4.5
zai-org
- - - 3.14 43 No No
Qwen3-Next-80B-A3B-Instruct
novita
0.15 1.5 131072 0.70 141 Yes No
Qwen3-Next-80B-A3B-Instruct
hyperbolic
cheapest fastest
0.3 0.3 262144 0.57 167 Yes No
Qwen3-Next-80B-A3B-Instruct
together
0.15 1.5 262144 0.63 145 Yes Yes
DeepSeek-R1-Distill-Qwen-7B
nscale
0.15 0.15 131072 0.84 66 No Yes
DeepSeek-V3.1-Terminus
novita
0.27 1 131072 1.22 28 No Yes
DeepSeek-V3-0324
novita
0.27 1.12 163840 1.15 30 Yes Yes
DeepSeek-V3-0324
nebius
fastest
0.75 2.25 32768 1.00 147 No No
DeepSeek-V3-0324
sambanova
3 4.5 131072 0.92 132 Yes Yes
DeepSeek-V3-0324
hyperbolic
1.25 1.25 163840 1.81 30 No No
DeepSeek-V3-0324
together
1.25 1.25 131072 1.23 37 Yes Yes
DeepSeek-V3-0324
fireworks-ai
cheapest
0.9 0.9 163840 0.84 54 Yes No
ERNIE-4.5-VL-424B-A47B-Base-PT
novita
0.42 1.25 123000 1.44 41 No No
Qwen3-4B-Thinking-2507
nscale
0.01 0.03 262144 1.10 90 Yes No
Qwen3-VL-235B-A22B-Thinking
novita
0.98 3.95 131072 0.97 37 No No
gemma-2-2b-it
nebius
0.02 0.06 8192 0.39 154 No Yes
Qwen2.5-Coder-32B-Instruct
nscale
cheapest
0.06 0.2 131072 1.01 27 No Yes
Qwen2.5-Coder-32B-Instruct
hyperbolic
fastest
0.2 0.2 32768 0.99 76 No No
Qwen2.5-Coder-32B-Instruct
together
- - - - - - -
Qwen2.5-Coder-32B-Instruct
scaleway
- - - 0.47 39 Yes No
Qwen3-235B-A22B
novita
0.2 0.8 40960 1.05 27 No No
Qwen3-235B-A22B
nscale
cheapest
0.2 0.6 32000 1.35 23 Yes Yes
Qwen3-235B-A22B
together
0.2 0.6 40960 0.55 38 Yes Yes
Qwen3-235B-A22B
fireworks-ai
fastest
0.22 0.88 131072 0.90 71 Yes No
Qwen3-32B
groq
0.29 0.59 131072 0.21 252 Yes No
Qwen3-32B
novita
0.1 0.45 40960 0.46 90 No No
Qwen3-32B
nebius
0.1 0.3 40960 0.43 46 Yes Yes
Qwen3-32B
cerebras
fastest
0.4 0.8 - 0.29 295 No No
Qwen3-32B
sambanova
0.4 0.8 32768 2.78 190 Yes Yes
Qwen3-32B
nscale
cheapest
0.08 0.25 40960 0.96 26 Yes Yes
Qwen3-Next-80B-A3B-Thinking
novita
0.15 1.5 131072 0.86 118 Yes No
Qwen3-Next-80B-A3B-Thinking
hyperbolic
cheapest
0.3 0.3 262144 0.60 176 Yes No
Qwen3-Next-80B-A3B-Thinking
together
fastest
0.15 1.5 262144 0.49 187 Yes Yes
Llama-4-Scout-17B-16E-Instruct
groq
fastest
0.11 0.34 131072 0.22 369 Yes No
Llama-4-Scout-17B-16E-Instruct
novita
0.1 0.5 131072 0.62 35 No No
Llama-4-Scout-17B-16E-Instruct
sambanova
- - - - - - -
Llama-4-Scout-17B-16E-Instruct
nscale
cheapest
0.09 0.29 890000 0.98 33 Yes Yes
Llama-4-Scout-17B-16E-Instruct
together
0.18 0.59 1048576 0.40 83 Yes Yes
Llama-4-Scout-17B-16E-Instruct
fireworks-ai
0.15 0.6 1048576 0.74 68 Yes No
Llama-4-Maverick-17B-128E-Instruct
groq
cheapest
0.2 0.6 131072 2.73 195 Yes No
Llama-4-Maverick-17B-128E-Instruct
sambanova
fastest
0.63 1.8 131072 1.45 337 Yes Yes
Llama-4-Maverick-17B-128E-Instruct
fireworks-ai
0.22 0.88 1048576 0.57 95 Yes No
DeepSeek-R1-0528
novita
0.7 2.5 163840 0.90 26 Yes No
DeepSeek-R1-0528
nebius
cheapest
0.8 2.4 163840 0.50 26 Yes Yes
DeepSeek-R1-0528
sambanova
fastest
5 7 131072 1.57 152 Yes Yes
DeepSeek-R1-0528
hyperbolic
3 3 163840 1.27 44 No No
DeepSeek-R1-0528
together
3 7 163840 1.07 40 No Yes
DeepSeek-R1-0528
fireworks-ai
3 8 163840 1.10 36 Yes No
DeepSeek-V3.1
novita
cheapest
0.27 1 131072 1.25 27 Yes Yes
DeepSeek-V3.1
together
fastest
0.6 1.7 131072 1.88 94 Yes No
DeepSeek-V3.1
fireworks-ai
- - 163840 0.98 72 Yes No
Qwen3-VL-235B-A22B-Instruct
novita
0.3 1.5 131072 0.93 41 Yes No
DeepSeek-R1-Distill-Qwen-14B
novita
cheapest fastest
0.15 0.15 32768 1.10 57 No No
DeepSeek-R1-Distill-Qwen-14B
nscale
0.2 0.2 131072 1.17 37 No No
Qwen3-Coder-480B-A35B-Instruct
novita
cheapest
0.29 1.2 262144 0.74 39 Yes Yes
Qwen3-Coder-480B-A35B-Instruct
nebius
0.4 1.8 262144 0.48 59 Yes Yes
Qwen3-Coder-480B-A35B-Instruct
cerebras
- - - - - - -
Qwen3-Coder-480B-A35B-Instruct
hyperbolic
fastest
2 2 262144 1.91 64 Yes No
Qwen3-Coder-480B-A35B-Instruct
together
2 2 262144 0.55 37 Yes Yes
Qwen3-Coder-480B-A35B-Instruct
fireworks-ai
0.45 1.8 262144 1.39 48 Yes No
Qwen2.5-Coder-7B-Instruct
nscale
0.01 0.03 131072 1.04 60 No Yes
NVIDIA-Nemotron-Nano-12B-v2
nebius
0.07 0.2 131072 0.36 124 No Yes
c4ai-command-a-03-2025
cohere
- - - 0.53 62 Yes No
Qwen3-30B-A3B-Thinking-2507
nebius
0.1 0.3 262144 0.46 109 Yes Yes
gemma-2-9b-it
groq
- - - - - - -
gemma-2-9b-it
nebius
0.03 0.09 8192 0.40 122 No Yes
command-a-translate-08-2025
cohere
- - - 0.39 63 Yes No
DeepSeek-R1-Distill-Llama-8B
novita
- - - - - - -
DeepSeek-R1-Distill-Llama-8B
nscale
0.05 0.05 131072 0.86 57 No Yes
Qwen3-30B-A3B
novita
cheapest
0.09 0.45 40960 1.06 103 No No
Qwen3-30B-A3B
fireworks-ai
fastest
0.15 0.6 131072 0.69 141 Yes No
Llama-3.1-70B-Instruct
fireworks-ai
0.9 0.9 131072 0.43 133 No No
DeepSeek-R1-Distill-Llama-70B
groq
- - - - - - -
DeepSeek-R1-Distill-Llama-70B
novita
0.8 0.8 8192 2.13 30 No Yes
DeepSeek-R1-Distill-Llama-70B
sambanova
fastest
0.7 1.4 131072 0.86 199 No No
DeepSeek-R1-Distill-Llama-70B
nscale
cheapest
0.75 0.75 131072 0.96 16 No No
DeepSeek-R1-Distill-Llama-70B
scaleway
- - - 0.73 21 No Yes
Qwen2.5-72B-Instruct
novita
cheapest
0.38 0.4 32000 0.64 25 Yes No
Qwen2.5-72B-Instruct
hyperbolic
0.4 0.4 131072 1.06 69 Yes No
Qwen2.5-72B-Instruct
together
fastest
1.2 1.2 131072 0.41 86 Yes Yes
Qwen2.5-72B-Instruct
fireworks-ai
- - - - - - -
Llama-4-Maverick-17B-128E-Instruct-FP8
novita
cheapest
0.17 0.85 1048576 0.78 57 No No
Llama-4-Maverick-17B-128E-Instruct-FP8
together
fastest
0.27 0.85 1048576 2.02 84 Yes Yes
Meta-Llama-3-70B-Instruct
groq
- - - - - - -
Meta-Llama-3-70B-Instruct
novita
0.51 0.74 8192 0.81 28 No No
Meta-Llama-3-70B-Instruct
hyperbolic
cheapest
0.4 0.4 8192 0.76 95 No No
Meta-Llama-3-70B-Instruct
together
fastest
0.88 0.88 8192 0.45 95 No Yes
Qwen2.5-VL-72B-Instruct
nebius
0.25 0.75 32000 0.45 35 No Yes
Qwen2.5-VL-72B-Instruct
hyperbolic
cheapest fastest
0.6 0.6 32768 0.54 43 No No
ERNIE-4.5-VL-28B-A3B-PT
novita
0.14 0.56 30000 0.86 79 No No
ERNIE-4.5-0.3B-PT
novita
- - 120000 1.22 98 No No
GLM-4.1V-9B-Thinking
novita
0.035 0.138 65536 0.71 70 No No
Baichuan-M2-32B
novita
0.07 0.07 131072 2.12 31 No Yes
Arch-Router-1.5B
hf-inference
- - - 0.22 76 No Yes
Llama-Guard-4-12B
groq
0.2 0.2 131072 0.27 7 No No
GLM-4.5V-FP8
zai-org
- - - 2.91 51 Yes No
GLM-4.6-FP8
zai-org
- - - 3.27 40 No No
AFM-4.5B
together
0.048 0.15 65536 0.57 188 No Yes
Qwen3-Coder-480B-A35B-Instruct-FP8
together
2 2 262144 0.81 37 Yes Yes
L3-8B-Stheno-v3.2
novita
0.05 0.05 8192 0.72 65 No No
L3-70B-Euryale-v2.1
novita
1.48 1.48 8192 0.79 41 No No
DeepSeek-Prover-V2-671B
novita
0.7 2.5 160000 0.58 27 No No
MiniMax-M1-80k
novita
0.55 2.2 1000000 1.25 56 Yes No
Hermes-4-70B
nebius
0.13 0.4 131072 0.43 64 No No
ERNIE-4.5-21B-A3B-PT
novita
0.07 0.28 120000 1.19 78 No No
Qwen3-VL-30B-A3B-Thinking
novita
0.2 1 131072 0.88 88 Yes Yes
QwQ-32B
groq
- - - - - - -
QwQ-32B
nebius
- - - - - - -
QwQ-32B
sambanova
- - - - - - -
QwQ-32B
nscale
cheapest
0.18 0.2 131072 1.07 27 Yes No
QwQ-32B
hyperbolic
fastest
0.4 0.4 131072 1.05 59 No No
QwQ-32B
fireworks-ai
- - - - - - -
Qwen2.5-Coder-3B-Instruct
nscale
0.01 0.03 32768 1.19 73 No Yes
c4ai-command-r7b-arabic-02-2025
cohere
- - - 0.42 1 No No
Apertus-70B-Instruct-2509
publicai
- - - 1.13 39 Yes Yes
aya-expanse-32b
cohere
- - - 0.27 42 No No
c4ai-command-r7b-12-2024
cohere
- - - 0.27 117 Yes No
Qwen3-235B-A22B-Thinking-2507
novita
0.3 3 131072 0.96 37 Yes No
Qwen3-235B-A22B-Thinking-2507
cerebras
- - - - - - -
Qwen3-235B-A22B-Thinking-2507
fireworks-ai
cheapest fastest
0.22 0.88 262144 0.89 53 Yes No
Qwen2.5-VL-32B-Instruct
fireworks-ai
0.22 0.88 128000 0.55 55 No No
Dobby-Unhinged-Llama-3.3-70B
fireworks-ai
0.9 0.9 131072 0.47 51 No No
GLM-4.5-Air-FP8
together
0.2 1.1 131072 0.67 120 Yes Yes
Hermes-4-405B
nebius
1 3 131072 0.40 31 No No
Llama-3.1-405B-Instruct
sambanova
- - - - - - -
Llama-3.1-405B-Instruct
fireworks-ai
3 3 131072 0.60 63 Yes No
Mixtral-8x22B-Instruct-v0.1
nscale
cheapest
1.2 1.2 65536 1.10 25 No Yes
Mixtral-8x22B-Instruct-v0.1
together
- - - - - - -
Mixtral-8x22B-Instruct-v0.1
fireworks-ai
fastest
1.2 1.2 65536 0.43 70 No No
cogito-v2-preview-deepseek-671B-MoE
together
1.25 1.25 163840 1.26 46 No Yes
cogito-v2-preview-llama-405B
together
3.5 3.5 32768 1.32 26 Yes Yes
cogito-v2-preview-llama-109B-MoE
together
0.18 0.59 32767 0.66 84 Yes Yes
cogito-v2-preview-llama-70B
together
0.88 0.88 32768 0.84 53 Yes Yes
command-a-reasoning-08-2025
cohere
- - - 0.24 89 Yes No
command-a-vision-07-2025
cohere
- - - 0.28 45 No No
aya-vision-32b
cohere
- - - 0.30 56 No No
c4ai-command-r-08-2024
cohere
- - - 0.35 55 Yes No
Llama-3.3-Swallow-70B-Instruct-v0.4
sambanova
0.6 1.2 131072 2.80 139 No Yes
Gemma-SEA-LION-v4-27B-IT
publicai
- - - 1.83 48 No Yes
DeepSeek-R1-Distill-Qwen-32B
novita
cheapest fastest
0.3 0.3 64000 0.69 38 No Yes
DeepSeek-R1-Distill-Qwen-32B
nscale
0.3 0.3 131072 1.28 25 No Yes
ERNIE-4.5-300B-A47B-Base-PT
novita
0.28 1.1 123000 1.24 30 No Yes
Hermes-3-Llama-3.1-70B
hyperbolic
0.4 0.4 12288 0.35 30 No No
GLM-4-32B-0414
novita
0.55 1.66 32000 1.07 37 No Yes
L3-8B-Lunaris-v1
novita
0.05 0.05 8192 0.71 71 No No
Hermes-2-Pro-Llama-3-8B
novita
0.14 0.14 8192 0.90 67 No No
WizardLM-2-8x22B
novita
0.62 0.62 65535 0.43 46 No No
Llama-3_1-Nemotron-Ultra-253B-v1
nebius
0.6 1.8 131072 0.45 39 No Yes
Qwen2.5-Coder-7B
nebius
0.03 0.09 32768 0.39 175 No Yes
Qwen3-235B-A22B-FP8
together
0.2 0.6 40960 0.52 38 Yes Yes
marin-8b-instruct
together
0.18 0.18 4096 0.37 172 No Yes