Mode VC LLM Token Economics & Model Release Velocity
Query empirical frontier LLM release velocity, inference token deflation, and enterprise AI spend.
我該用這個嗎
品質與安全性
發現項目(3)
- HIGH
- MEDIUM在 calc_llm_inference_cost 中
- LOW在 calc_llm_inference_cost 中
根據工具定義與協定合規性的自動化分析。
上下文成本
這是每次將伺服器的工具載入模型上下文時所消耗的約略 token 數量。數量越高,可用於其他工作的注意力就越少。
安裝
一鍵安裝
將以下內容加入你的 `claude_desktop_config.json` 檔案:
{
"mcpServers": {
"llm-economics": {
"url": "https://modevc.com/api/mcp/llm-economics"
}
}
}遠端端點
https://modevc.com/api/mcp/llm-economicsstreamable-http它能做什麼
工具清單
工具(4)
🟢get_venture_chart_of_the_day(limit)
Returns the most recent charts from Mode VC's Chart of the Day archive (one new chart per day, mostly AI model, adoption, developer-package and pricing data), newest first, each with its claim, takeaway, URL and cited sources.
輸入結構描述
{
"type": "object",
"properties": {
"limit": {
"type": "integer",
"minimum": 1,
"maximum": 10,
"description": "Number of recent charts. Default 3."
}
},
"additionalProperties": false
}🟢search_charts(query, limit)
Searches every chart in the Mode VC Chart of the Day archive by keyword over title, claim, takeaway and keywords. Returns claims, takeaways, canonical URLs, markdown endpoints and cited sources. The archive grows by one chart a day; the response states its current size.
輸入結構描述
{
"type": "object",
"properties": {
"query": {
"type": "string",
"description": "Search term, e.g. \"pricing\", \"claude\", \"inference\", \"sdk\""
},
"limit": {
"type": "integer",
"minimum": 1,
"maximum": 100,
"description": "Maximum charts to return. Default 10."
}
},
"additionalProperties": false
}🟢get_chart_data(slug)
Returns one chart from the Chart of the Day archive: the data table behind it (rows as published), claim, takeaway, note and the sources it cites. Look up by slug or title. Data is as published on the chart page; follow source_urls for the original data.
輸入結構描述
{
"type": "object",
"properties": {
"slug": {
"type": "string",
"description": "Chart slug or title, e.g. \"pre-seed-valuation-dispersion\", \"ai-inference-cost-comparison\""
}
},
"required": [
"slug"
],
"additionalProperties": false
}🟢calc_llm_inference_cost(monthly_active_users, requests_per_user, input_tokens_per_request, output_tokens_per_request, model, ...)
Estimates monthly LLM spend, cost per user and per request, prompt-cache savings and gross margin for a usage profile on one of 8 models, compares all 8 models, and prices an 80/20 hybrid of DeepSeek and Claude 3.7 Sonnet. Same math and the same fixed price table as https://modevc.com/tools/llm-inference-calculator. The prices are that page's illustrative table, not a live price feed.
輸入結構描述
{
"type": "object",
"properties": {
"monthly_active_users": {
"type": "number",
"exclusiveMinimum": 0,
"minimum": 1,
"zeroNote": "the web calculator replaces 0 with its default",
"description": "Web default 5,000."
},
"requests_per_user": {
"type": "number",
"exclusiveMinimum": 0,
"minimum": 1,
"zeroNote": "the web calculator replaces 0 with its default",
"description": "Requests per user per month. Web default 40."
},
"input_tokens_per_request": {
"type": "number",
"exclusiveMinimum": 0,
"minimum": 1,
"zeroNote": "the web calculator replaces 0 with its default",
"description": "Web default 1,200."
},
"output_tokens_per_request": {
"type": "number",
"exclusiveMinimum": 0,
"minimum": 1,
"zeroNote": "the web calculator replaces 0 with its default",
"description": "Web default 400."
},
"model": {
"type": "string",
"enum": [
"claude-3-7-sonnet",
"gpt-4o",
"gemini-2-5-pro",
"deepseek-v4",
"claude-3-5-haiku",
"gpt-4o-mini",
"gemini-2-5-flash",
"groq-lpu-qwen"
],
"description": "Primary model. Web default claude-3-7-sonnet."
},
"cache_hit_pct": {
"type": "number",
"minimum": 0,
"maximum": 100,
"description": "Share of input tokens served from the prompt cache, in percent. Web default 40."
},
"price_per_user_usd": {
"type": "number",
"exclusiveMinimum": 0,
"zeroNote": "the web calculator replaces 0 with its default",
"description": "Monthly revenue per user in USD. Web default 20."
}
},
"additionalProperties": false
}社群
證據