Unverified50% confidenceFactExact time
OpenRouter data for the past 30 days shows that the old version of DeepSeek Flash processed about 122 trillion prompt tokens and about 2.2 trillion completion tokens, a ratio of about 55:1, and most prompts were cache hits.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/18/2026
First Seen
Valid until: 11/1/2026
Sources
DeepSeek New Version Hands-On: Same Price, Costs Slashed by 60%
bilibili枫叶边城10/18/2026
Related Entities
Related Claims
UnverifiedDeepSeek's flagship version has approximately 1.6 trillion total parameters and 49 billion activated parameters; the lightweight Flash version has 284 billion total parameters and 13 billion activated parameters, supports a context window of 1 million tokens, and is open-sourced under the MIT license.72% similarUnverifiedThe developer's Claude Code logs over three days show a total of about 65,000 requests, 12.6 billion tokens processed, and only 14 million tokens of actual output.66% similarUnverifiedThe author developed GFX1151 Engine primarily with assistance from large models such as Kimi and ChatGPT; after independently reaching approximately 600 tokens/s, further improvement proved difficult, and the speed was ultimately raised to 800–1,000 tokens/s by reverse-engineering and referencing the Hologen Flash Server approach.64% similarUnverifiedAt the beginning of the year, OpenCode had approximately 650,000 monthly active users.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1059846API
curl https://kongchang.com/api/v1/knowledge/claims/1059846MCP
get_claim(id=1059846)