Unverified50% confidenceFactExact time
Llama 3.3 70B Instruct on DeepInfra's FP8 serverless endpoint has an input price as low as $0.12/1M.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/19/2026
First Seen
Valid until: 11/2/2026
Sources
Related Entities
Related Claims
UnverifiedDeepSeek V3's activated parameters are compressed to 3 to 5 billion, with a total parameter count of 671B.68% similarUnverifiedA single DGX Spark with 128GB of unified memory and 273GB/s of bandwidth cannot hold DeepSeek 1017's expert weights, so the community typically forms a 512GB cluster with 4 units; the DGX Spark's price has risen from 3999 dollars to about 6950 dollars, and four units cost about 27800 dollars excluding networking.65% similarUnverifiedRunPod serverless and dedicated GPU cloud service prices range from $0.20/hour to $2.69/hour, supporting vLLM template deployment of Llama 3.3 70B, DeepSeek R1 Distill, and Qwen 2.5 Coder 32B.64% similarUnverifiedDeepSeek V4 Pro 0423's channel input price is 40% lower than the official input price, and its output price is 51% lower.64% similarUnverifiedThe KV cache overhead per token for Llama 3.1 8B is approximately 128 KB.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1072608API
curl https://kongchang.com/api/v1/knowledge/claims/1072608MCP
get_claim(id=1072608)