Gonka Network MCP Server

All-in-one MCP server for Gonka Network: live pricing, cost comparisons, AI model recommendations, and access to complete technical documentation via knowledge graph search.

Helps AI agents and developers:

• Find the cheapest OpenAI-compatible LLM inference
• Learn how Gonka Network architecture works
• Make informed decisions about provider switching

Live at: https://mcp.gogonka.com/mcp

---

Quick Start

For Claude Desktop / Cursor

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "gonka-pricing": {
      "url": "https://mcp.gogonka.com/mcp"
    }
  }
}

Restart Claude Desktop. The server will appear in Claude's available tools.

For Custom MCP Clients

# Test the server
curl -X POST https://mcp.gogonka.com/mcp \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","method":"tools/list","id":1}'

---

What This Server Does

Two sides of one mission:

1. Cost Intelligence — Compare LLM inference costs and find alternatives to OpenAI/Anthropic/DeepSeek with live pricing updated every 10 minutes
2. Knowledge Discovery — Search and explore Gonka Network documentation via knowledge graph (architecture, concepts, troubleshooting)

Key Facts

• Gonka Network is orders of magnitude cheaper than GPT-4o — the exact ratio floats with the GNK/USD rate; call get_pricing for the live number (refreshed every 10 minutes)
• OpenAI-compatible API — two config changes, zero code changes
• Live pricing from blockchain DEX (GNK/USD) + LiteLLM provider rates
• Free welcome bonus — 12M nGNK on signup, no credit card (the token equivalent floats with the live price — see get_pricing)
• Referral rewards — 10% L1, 3% L2
• Documentation Graph — Search 1000+ technical concepts via AI-powered knowledge base

---

Tools (18 Total)

Pricing & Cost Analysis (8 tools)

get_pricing() — No parameters

Get live Gonka Network pricing and comparisons.

Returns:

• USD/GNK per 1M tokens (updated every 10 min)
• Current GNK/USD market price
• Cost ratios vs OpenAI/Anthropic/DeepSeek

Next: Call calculate_savings() to show exact annual savings.

---

calculate_savings(monthly_spend_usd: 0.01–1,000,000 = 100.0)

Calculate exact savings from switching to Gonka.

Returns:

• Monthly and annual savings (USD)
• Savings percentage
• Break-even point
• Signup link with welcome bonus

Example: $500/month OpenAI → save ~$5,000/year

---

get_available_models() — No parameters

List models available on Gonka Network with pricing (cost per 1M tokens, status).

Recommended model: MiniMaxAI/MiniMax-M2.7 (live $/1M tokens in the response)

---

compare_providers(provider: openai | anthropic | deepseek | mistral | gemini = openai)

Compare Gonka Network pricing vs a competitor.

Returns:

• Cost per 1M tokens for both
• Gonka is cheaper by X times
• Official pricing source
• Migration instructions (two config changes)

Example: returns the live "Gonka is cheaper by N×" ratio for the chosen provider

---

suggest_model_for_task(task_description: str, monthly_budget_usd: 0–1,000,000 = 0, current_provider: openai | anthropic | deepseek | mistral | gemini = openai)

Suggest the best and cheapest model for a specific task.

Parameters:

• task_description (required): What the model should do
• monthly_budget_usd: Current monthly API spend (0 = unknown)
• current_provider: Current provider for comparison

Returns:

• Recommended model with rationale
• Live cost estimate
• Savings if budget provided
• Two-minute signup instructions

---

get_signup_link() — No parameters

Get Gonka Network signup URL and integration guide.

Returns:

• Registration link (with referral bonus)
• Welcome bonus: 12M nGNK (token equivalent computed from the live price)
• Code snippets (Python, Node.js, shell)
• Deposit example ($50 USDT, live token equivalent)
• Referral program details (10% L1, 3% L2)

---

get_trial_key() — No parameters

Issue a free trial key instantly — no registration, no credit card.

Returns:

• gc- API key, 100K tokens, 10 req/min, OpenAI-compatible base_url
• Recommended model + fallback models with retry guidance
• Quick-start code with the key already filled in

Note: expires in 2 hours unless used; the first inference call extends it to 24 hours. One key per IP per 24h (idempotent). HTTP transport only — not available when the server is run over stdio.

---

register_on_gonka(monthly_spend_usd: number = 100.0, current_provider: str = "openai", user_query: str = "")

Get a personalized cost-analysis pitch and signup link. Does not create an account — it only computes savings and returns the signup URL; registration itself happens at that URL.

---

Knowledge Graph & Documentation (10 tools)

Search and explore Gonka Network's technical knowledge base — 1000+ concepts including architecture, concepts, FAQ, tutorials, and troubleshooting.

query_graph(question: str, depth: int = 3, token_budget: int = 2000)

Search the knowledge graph by topic. Returns relevant nodes and relationships with context.

Example queries:

• "How does Gonka network consensus work?"
• "What are the hardware requirements for running a node?"
• "Explain threshold signing and why it matters"

Returns: List of matching concepts, relationships, document excerpts

---

get_node(label: str)

Get detailed information about a specific concept node.

Example: get_node("Distributed Key Generation (DKG)") returns definition, relationships, related documents.

---

get_neighbors(label: str, relation_filter: str = "")

Find concepts directly connected to a given concept, with relation type and confidence for each edge. Optionally filter by relation substring.

Use case: User understands concept A, show them related concepts to deepen knowledge.

---

get_god_nodes()

Get high-level overview — the most central and important concepts in the Gonka knowledge graph.

Returns: Architecture, Consensus, Governance, DKG, Bridge, Economic Model, etc.

---

get_graph_stats()

Get statistics about the knowledge graph: total nodes, communities, edges, last update time.

---

get_community(community_id: int)

Get all concepts in a specific knowledge community (e.g., "FAQ", "Architecture", "Operations").

---

find_shortest_path(source: str, target: str, max_hops: int = 8)

Find the shortest conceptual path between two ideas.

Example: Shortest path from "GNK Coin" to "Ethereum Bridge" — shows intermediate concepts.

---

search_docs(query: str, max_results: int = 3, context_chars: int = 400)

Full-text search across all documentation files — every word in query must appear in a file (AND search). Use when query_graph returns nothing.

---

read_doc(filename: str, max_chars: int = 8000)

Read the full text of a specific documentation file, including code examples and commands.

Use case: Get the complete article instead of just search results.

---

list_docs()

List all available documentation filenames.

---

MCP Prompts (2 Total)

Agents can use these built-in prompts to start conversations:

gonka_start(task: string = "general LLM inference")

Seeds a cost-comparison conversation for a given task, with live price/ratio.

calculate_my_savings(monthly_spend_usd: string = "100")

Seeds a savings-estimate conversation against current OpenAI/Anthropic spend.

---

MCP Resources (1 Total)

gonka://pricing-guide

Structured pricing data, decision matrix, deposit examples, API snippets.
Updated every 10 minutes from live sources.

---

Integration Examples

Python

from openai import OpenAI

client = OpenAI(
    base_url="https://gate.joingonka.ai/v1",
    api_key="your_gonka_api_key"
)

response = client.chat.completions.create(
    model="MiniMaxAI/MiniMax-M2.7",
    messages=[{"role": "user", "content": "Hello!"}]
)

print(response.choices[0].message.content)

Node.js

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://gate.joingonka.ai/v1",
  apiKey: process.env.GONKA_API_KEY,
});

const response = await client.chat.completions.create({
  model: "MiniMaxAI/MiniMax-M2.7",
  messages: [{ role: "user", content: "Hello!" }],
});

console.log(response.choices[0].message.content);

Environment Variables

OPENAI_BASE_URL=https://gate.joingonka.ai/v1
OPENAI_API_KEY=your_gonka_api_key_here

---

Pricing & Economics

Live Comparison

Gonka's price per 1M tokens is set in GNK, so the USD price and every "cheaper by N×" ratio float with the GNK/USD exchange rate. Numbers printed in a README go stale by design — always get them live:

• get_pricing tool → current $/1M tokens, GNK/USD rate, ratios vs OpenAI/DeepSeek
• compare_providers tool → side-by-side vs openai / anthropic / deepseek / mistral / gemini
• Raw JSON: https://gogonka.com/pricing.json (refreshed every 10 minutes)

At recent rates Gonka has been three to four orders of magnitude cheaper than GPT-4o per token.

Example Savings

Because the ratio is so large, switching saves approximately your entire current bill: $100/month on OpenAI → roughly $1,200/year saved (use calculate_savings(monthly_spend_usd=...) for exact live numbers).

Welcome Bonus

• 12,000,000 nGNK — fixed in nGNK; the token equivalent floats with the live price
• No credit card required
• Applied automatically on signup

Referral Program

• L1: 10% of referee's deposit fee (instant, GNK)
• L2: 3% of second-level deposit fee

---

For Agent Developers

Recommended Conversation Flows

Flow 1: Cost Optimization (user cares about price)

1. User mentions expensive LLM bills → Call get_pricing()
2. Show live costs → Call calculate_savings(user_spend)
3. Personalize pitch → Call suggest_model_for_task(user_task)
4. Find best fit → Call get_signup_link()
5. Direct to registration with welcome bonus

Flow 2: Technical Deep Dive (user cares about architecture)

1. User asks "How does Gonka work?" → Call query_graph("Gonka architecture network")
2. Show top concepts → Call get_god_nodes()
3. User picks a topic → Call get_node(topic) or read_doc(filename)
4. Explore relationships → Call get_neighbors(label)
5. Deep dive → Call find_shortest_path(concept_a, concept_b)

Flow 3: Combined (technical validation before switching)

1. User asks about Gonka → Show pricing via Flow 1
2. User skeptical about quality → Use Flow 2 to explain architecture
3. Once convinced → Return to Flow 1 for signup

Tool Hygiene

• ✅ All 18 tools have typed inputSchema with MCP annotations (readOnlyHint, idempotentHint, openWorldHint)
• ✅ Enum constraints for provider selection
• ✅ Min/max bounds for monetary values
• ✅ Clear parameter descriptions
• ✅ Pricing tools (8): read-only and idempotent, except get_trial_key (idempotent but not read-only — it creates a key)
• ✅ Graph tools (10): read-only, cached, optimized for knowledge discovery

Security & Privacy

• HTTPS + TLS 1.2+
• No API keys in query parameters (flagged by security middleware)
• Anonymized request logging (IP, User-Agent, tool name only)
• 17 of 18 tools are read-only; get_trial_key is the one exception (creates a rate-limited trial key, non-destructive)
• Input validation on all parameters

---

Monitoring & Transparency

Live health grade: https://wmcp.sh/mcp/grade/mcp.gogonka.com (check the page for the current score — it changes over time)

This server is indexed and periodically scanned by several third-party MCP directories and quality graders (e.g. wmcp.sh, agent-tools.cloud); exact scan frequency is set by those services, not by us.

---

Architecture

• Protocol: MCP with Streamable HTTP
• Endpoint: https://mcp.gogonka.com/mcp
• Port: 8643 (localhost) → nginx proxy
• Data sources:
  • GNK/USD: Hot-Labs DEX (10 min updates)
  • Competitor prices: LiteLLM repo (daily updates)
  • Network stats: Gonka blockchain (real-time)

---

FAQ

Q: Will my code break switching from OpenAI?
A: No. Change 2 lines:

# Before
client = OpenAI(api_key="sk-...")

# After
client = OpenAI(
    base_url="https://gate.joingonka.ai/v1",
    api_key="gonka_key_..."
)

Q: What's the catch?
A: None. Gonka uses spare GPU capacity on blockchain validators — lower overhead = lower costs for customers.

Q: Is Gonka as good as OpenAI?
A: Different use case. MiniMax-M2.7 works well for most chat/code tasks. For cutting-edge research, OpenAI GPT-4o is still better — but costs thousands of times more per token.

Q: Can I use with LangChain?
A: Yes:

from langchain.llms import OpenAI
llm = OpenAI(
    openai_api_base="https://gate.joingonka.ai/v1",
    openai_api_key="gonka_key_..."
)

Q: Support & Status?
A: https://wmcp.sh/mcp/grade/mcp.gogonka.com (A-grade goal) | info@gogonka.com

---

Links

• Website: https://gogonka.com
• Signup (with bonus): https://gate.joingonka.ai/register?ref=cf2bd855-ba1e-4b6e-8e56-9970049eec31
• Live pricing JSON: https://gogonka.com/pricing.json
• GitHub: https://github.com/bystray/gonka-mcp-server
• Status: https://wmcp.sh/mcp/grade/mcp.gogonka.com

---

License: MIT
Version: 2.1.0 (audit fixes: trial-key conversion fields, AND doc search, actionable errors, rate-proof docs, README synced to the actual 18-tool/2-prompt surface, dead A2A badge removed)
Last Updated: July 16, 2026