DeepSeek V4 Pro 1.6T
8K/1K
B200
vLLM · FP4
MI355X
$0.10453% cheaper than B200
ATOM¹ · FP4
B300
$0.18218% cheaper than B200
vLLM · FP4
GB200 NVL72
$0.225About the same cost as B200
Dynamo vLLM · FP4
GB300 NVL72
—
no exact @30 result
Inference Cost per Million Tokens
Hyperscaler cost ↓ Lower is better Source: InferenceX & SemiAnalysis Market July 2026 AI Cloud TCO Model
| Model · Scenario | B200 · Reference | MI355X | B300 | GB200 NVL72 | GB300 NVL72 |
|---|---|---|---|---|---|
DeepSeek V4 Pro 1.6T8K/1K | vLLM · FP4 | $0.10453% cheaper than B200 ATOM¹ · FP4 | $0.18218% cheaper than B200 vLLM · FP4 | $0.225About the same cost as B200 Dynamo vLLM · FP4 | — no exact @30 result |
DeepSeek V4 Pro 1.6TLong Context Multi-Turn Realistic Agentic Scenario (AgentX) | vLLM · FP4 | $0.071218% more expensive than B200 vLLM · FP4 | $0.0248% more expensive than B200 vLLM · FP4 | — no data for this scenario | — no exact @30 result |
Kimi K3 2.8TLong Context Multi-Turn Realistic Agentic Scenario (AgentX) | Dynamo vLLM · FP4 · STP | — no data for this scenario | $0.13587% cheaper than B200 vLLM · FP4 | $0.42959% cheaper than B200 Dynamo vLLM · FP4 | — no data for this scenario |
MiniMax M3 428B8K/1K | vLLM · FP8 | $0.04719% cheaper than B200 ATOM¹ · FP4 | $0.04620% cheaper than B200 vLLM · FP4 | $0.08649% more expensive than B200 Dynamo vLLM · FP8 · STP | — no exact @30 result |
MiniMax M3 428BLong Context Multi-Turn Realistic Agentic Scenario (AgentX) | — no exact @30 result | — no data for this scenario | — no exact @30 result | — no data for this scenario | — no data for this scenario |
GLM5.2Long Context Multi-Turn Realistic Agentic Scenario (AgentX) | — no exact @30 result | — no data for this scenario | — no exact @30 result | — no data for this scenario | — no data for this scenario |
Qwen3.5 397B8K/1K | TRTLLM · FP4 | $0.05769% more expensive than B200 SGLang · FP4 | $0.069107% more expensive than B200 SGLang · FP8 | — no exact @30 result | — no exact @30 result |
Qwen3.5 397BLong Context Multi-Turn Realistic Agentic Scenario (AgentX) | — no exact @30 result | — no data for this scenario | — no exact @30 result | — no data for this scenario | $0.019No B200 baseline to compare against Dynamo SGLang · FP4 |
DeepSeek R1 0528 671BMaintenance8K/1K | TRTLLM · FP4 | $0.123131% more expensive than B200 ATOM¹ · FP4 | $0.139160% more expensive than B200 Dynamo TRTLLM · FP4 | $0.10189% more expensive than B200 Dynamo TRTLLM · FP4 · STP | — no exact @30 result |
Kimi K2.5/2.6/2.7-Code 1TDeprecated8K/1K | vLLM · FP4 · STP | $0.0947% cheaper than B200 ATOM¹ · FP4 · STP | $0.12422% more expensive than B200 vLLM · FP4 · STP | — no exact @30 result | — no exact @30 result |
GLM5/5.1 744BDeprecated8K/1K | SGLang · FP4 | $0.283115% more expensive than B200 SGLang · FP8 | $0.16324% more expensive than B200 SGLang · FP4 | — no exact @30 result | — no exact @30 result |
gpt-oss 120BDeprecated8K/1K | TRTLLM · FP4 · STP | — no exact @30 result | — no data for this scenario | — no exact @30 result | — no data for this scenario |
MiniMax M2.5/2.7 230BDeprecated8K/1K | vLLM · FP4 · STP | $0.03622% more expensive than B200 ATOM¹ · FP4 · STP | $0.03725% more expensive than B200 vLLM · FP4 · STP | — no exact @30 result | $0.04245% more expensive than B200 Dynamo vLLM · FP4 · STP |
Llama 3.3 70B InstructDeprecated8K/1K | TRTLLM · FP4 · STP | $0.07864% more expensive than B200 vLLM · FP4 · STP | — no data for this scenario | — no data for this scenario | — no data for this scenario |
- View details
DeepSeek V4 Pro 1.6T
Long Context Multi-Turn Realistic Agentic Scenario (AgentX)
View detailsB200vLLM · FP4MI355X$0.071218% more expensive than B200vLLM · FP4B300$0.0248% more expensive than B200vLLM · FP4GB200 NVL72—no data for this scenarioGB300 NVL72—no exact @30 resultKimi K3 2.8T
Long Context Multi-Turn Realistic Agentic Scenario (AgentX)
View detailsB200Dynamo vLLM · FP4STPMI355X—no data for this scenarioB300$0.13587% cheaper than B200vLLM · FP4GB200 NVL72$0.42959% cheaper than B200Dynamo vLLM · FP4GB300 NVL72—no data for this scenarioMiniMax M3 428B
8K/1K
View detailsB200vLLM · FP8MI355X$0.04719% cheaper than B200ATOM¹ · FP4B300$0.04620% cheaper than B200vLLM · FP4GB200 NVL72$0.08649% more expensive than B200Dynamo vLLM · FP8STPGB300 NVL72—no exact @30 resultMiniMax M3 428B
Long Context Multi-Turn Realistic Agentic Scenario (AgentX)
View detailsB200—no exact @30 resultMI355X—no data for this scenarioB300—no exact @30 resultGB200 NVL72—no data for this scenarioGB300 NVL72—no data for this scenarioGLM5.2
Long Context Multi-Turn Realistic Agentic Scenario (AgentX)
View detailsB200—no exact @30 resultMI355X—no data for this scenarioB300—no exact @30 resultGB200 NVL72—no data for this scenarioGB300 NVL72—no data for this scenarioQwen3.5 397B
8K/1K
View detailsB200TRTLLM · FP4MI355X$0.05769% more expensive than B200SGLang · FP4B300$0.069107% more expensive than B200SGLang · FP8GB200 NVL72—no exact @30 resultGB300 NVL72—no exact @30 resultQwen3.5 397B
Long Context Multi-Turn Realistic Agentic Scenario (AgentX)
View detailsB200—no exact @30 resultMI355X—no data for this scenarioB300—no exact @30 resultGB200 NVL72—no data for this scenarioGB300 NVL72$0.019No B200 baseline to compare againstDynamo SGLang · FP4DeepSeek R1 0528 671BMaintenance
8K/1K
View detailsB200TRTLLM · FP4MI355X$0.123131% more expensive than B200ATOM¹ · FP4B300$0.139160% more expensive than B200Dynamo TRTLLM · FP4GB200 NVL72$0.10189% more expensive than B200Dynamo TRTLLM · FP4STPGB300 NVL72—no exact @30 resultKimi K2.5/2.6/2.7-Code 1TDeprecated
8K/1K
View detailsB200vLLM · FP4STPMI355X$0.0947% cheaper than B200ATOM¹ · FP4STPB300$0.12422% more expensive than B200vLLM · FP4STPGB200 NVL72—no exact @30 resultGB300 NVL72—no exact @30 resultGLM5/5.1 744BDeprecated
8K/1K
View detailsB200SGLang · FP4MI355X$0.283115% more expensive than B200SGLang · FP8B300$0.16324% more expensive than B200SGLang · FP4GB200 NVL72—no exact @30 resultGB300 NVL72—no exact @30 resultgpt-oss 120BDeprecated
8K/1K
View detailsB200TRTLLM · FP4STPMI355X—no exact @30 resultB300—no data for this scenarioGB200 NVL72—no exact @30 resultGB300 NVL72—no data for this scenarioMiniMax M2.5/2.7 230BDeprecated
8K/1K
View detailsB200vLLM · FP4STPMI355X$0.03622% more expensive than B200ATOM¹ · FP4STPB300$0.03725% more expensive than B200vLLM · FP4STPGB200 NVL72—no exact @30 resultGB300 NVL72$0.04245% more expensive than B200Dynamo vLLM · FP4STPLlama 3.3 70B InstructDeprecated
8K/1K
View detailsB200TRTLLM · FP4STPMI355X$0.07864% more expensive than B200vLLM · FP4STPB300—no data for this scenarioGB200 NVL72—no data for this scenarioGB300 NVL72—no data for this scenario
— = no result. ∞ = B200 baseline unavailable.
If a chip does not have FP4 spec decoding available, the next best available configuration is used.