COCKPIT / OVERVIEW
1658 CODING SCORE (QWEN-3.8-MAX-0902) DOCS → UPGRADE TIER
Welcome to MinTok Developer Cockpit, adstim!
Your workspace is active with 1,049 monthly credits. Start by generating an API key or running a test compilation.
AVAILABLE CREDITS LIVE
$10.49 USD
1049.00 cr balance + TOP UP
ACTIVE SUBSCRIPTION PRO
MinTok Pro
1049 cr / mo MANAGE →
Avoided Tokens ~81% AVOIDANCE
315300
79000 compiled tokens
ACTIVE API KEYS + NEW KEY
0
CLI Primary Key VIEW ALL →

Interactive Testing Playground & AST Compiler

Billing & Financial Ledger → 
# 1. Point your agent or OpenAI client to MinTok edge compiler:
export OPENAI_API_BASE="https://mintok.adstim.net/v1"
export OPENAI_API_KEY="mtk_live_your_key_here"

# 2. Run coding agent with MinTok 1 Max:
mintok /plan "Refactor authentication layer to enforce host-only session cookies"

Available Models & Routing Engine

4 models deployed with instant sub-millisecond AST context virtualization.

VIEW LMSYS LEADERBOARD →
Frontier Reasoning 1658 ELO
MinTok 1 Max

Multi-file refactoring, deep architectural planning, SWE-bench verified tasks

Context: 131,072 $0.00436 / task
Balanced Workhorse 1638 ELO
MinTok 1 Pro

Full-stack feature delivery, test suite generation, PR review

Context: 65,536 $0.00084 / task
Ultra-Fast Agile 1615 ELO
MinTok 1 Flash

Inline autocomplete, fast diff patches, git commit messages

Context: 32,768 $0.00051 / task
Predictive SEO Engine
MinTok 1 Keywords

Search Volume, CPC, SERP Competition & Expected Value (eV) prediction

Context: Sub-1ms Edge V8 $0.00015 / task
3-LAYER PRODUCTION MODEL-LEVEL PROOF SMOKING GUN: BACKEND PROMPT TOKENS

Deployed Model Verification: Claimed vs. Verified Fleet Status

Empirical verification proving the underlying deployed inference engines execute on compressed AST IR rather than original context tokens. All measurements backed by SHA-256 execution records and raw provider billing receipts.

FLEET JSON API →
MODEL UNDERLYING ENGINE CLAIMED BACKEND MEASURED REDUCTION STATUS AUDIT PROOF
mintok-1-flash
Generative AST Slicing
 MinTok Sliced IR Engine v2 4.5× 4.5× (240 / 1080 tok) -77.8% VERIFIED Proof A/B Test
mintok-1-pro
Generative AST Slicing
 MinTok Context Virtualizer v2 4.8× 4.8× (225 / 1080 tok) -79.2% VERIFIED Proof A/B Test
mintok-1-max
Generative AST Slicing
 kimi-k3-thinking (RunPod vLLM) 5.2× 5.2× (208 / 1080 tok) -80.8% VERIFIED Proof A/B Test
mintok-1-keywords
Specialized Tabular Avoidance
 LightGBM-LambdaRank-v2 (Decision Forest) 8.5× 8.5× (0 / 1250 LLM tok) -100.0% VERIFIED Proof A/B Test
Specialized Denominator Definition (mintok-1-keywords): Tabular models consume 0 generative GPU LLM tokens. The 8.5× avoidance ratio is measured against the 1,250 frontier reasoning tokens required by GPT-4o / Claude to classify SERP query intent and competition. Execution latency is 2.85 ms ($0.00015 / query).

Recent Agent Task Runs

Telemetry from autonomous coding sessions.

ALL RUNS →
Task ID Model Goal / Directive Status Compiled / Avoided Credits Time
task_17ad3d04 mintok-1-flash Fix race condition in WebSocket real-time event broadcaster COMPLETED 9500 / 42100 0.0028 cr Oct 03, 08:26
task_41eb46d1 mintok-1-pro Implement resilient database connection pooling with exponential backoff COMPLETED 20600 / 89200 0.0084 cr Oct 03, 08:26
task_1444dcd9 mintok-1-max Refactor monolithic authentication layer into modular JWT handlers COMPLETED 48900 / 184000 0.0436 cr Oct 03, 08:26