Coding Plan
A weekly AI coding model gateway for the Maximo Syntax CLI, Codex, Kilo Code, OpenClaw, IDE agents, and OpenAI SDKs. Start free with Maximo Atlas 1.2 and Atlas 1.3 Preview plus free 24-hour GLM-5.3 Flash and Qwen3.8 Flash launch windows, then unlock 21, 32, or 37 models from $3 per week. A paid Workspace Pro plan includes four weeks of Coding Plus, and Workspace Max includes four weeks of Coding Pro for the account that paid; the promotion is not shared with team members. Every model keeps its full native context.
Start in three steps
1) Open the Coding Plan dashboard and choose Free, Plus, Pro, or Max. Free is enabled automatically and needs no card. 2) Pick your tool — the Maximo Syntax CLI links to your plan with no API key, or create a live mtb_live_ API key with the ai.coding scope for any other client. 3) Copy the guide for your tool and select any model included in your tier. Chat Completions and Responses share one key, one model catalog, and the same usage pools. A Workspace Pro or Max promotion is automatically attached to the account that paid for the workspace plan, not the team.
Fastest start: the Maximo Syntax CLI
Install the CLI, run maximo, and choose the MyTabulon sign-in option. Your Coding Plan links automatically — no API key or config file needed.
npm install -g @maximoai/maximo-syntax-cli
maximoAny OpenAI-compatible client
Or use a live key with the ai.coding scope and an OpenAI-compatible client:
from openai import OpenAI
client = OpenAI(
base_url="https://api.mytabulon.com/v1",
api_key="mtb_live_...", # live key with the ai.coding scope
)
response = client.chat.completions.create(
model="maximo-atlas-1.2",
messages=[
{"role": "system", "content": "You are my pair programmer."},
{"role": "user", "content": "Add retry logic to this fetch call..."},
],
tools=[{
"type": "function",
"function": {
"name": "read_file",
"description": "Read a file from the workspace",
"parameters": {
"type": "object",
"properties": {"path": {"type": "string"}},
"required": ["path"],
},
},
}],
stream=True,
)Plans, model access, and concurrency
Coding Free includes Maximo Atlas 1.2, Atlas 1.3, and free 24-hour GLM-5.3 Flash and Qwen3.8 Flash launch windows with 2 concurrent requests. Atlas 1.1 is legacy and leaves the Maximo AI provider on August 21, 2026; new Coding Plan references resolve to Atlas 1.2. Atlas 1.2 public API pricing is normally $0.55/M cache-miss input, $0.05/M cached input, and $1.50/M output, with an automatic 80% sale at $0.11/$0.01/$0.30 per million from August 17 through August 31, 2026 UTC before standard rates return. Maximo Atlas 1.3 (maximo-atlas-1.3) is generally available on every Coding Plan tier at $0.20/M input, $0.03/M cached input, and $1.00/M output; the August 17–31, 2026 UTC launch sale at $0.11/$0.01/$0.30 per million has ended. Coding Plus is $3/week or ₦3,000/week, preserves the original Coding Plan usage pools, includes 21 models, and allows 10 concurrent requests. Coding Pro is $9/week or ₦9,000/week, includes 32 models, and allows 25 concurrent requests. Coding Max is $15/week or ₦15,000/week, includes all 37 models, and allows 50 concurrent requests. GLM-5.3 Flash (z-ai/glm-5.3-flash) — the model revealed behind Ox Alpha — is free on every tier for the first 24 hours after its August 26, 2026 launch, then requires Coding Plus+ at $0.075/M input, $0.015/M cached input, and $0.25/M output with a 1M-token context window, 131,072 max output, text/image input, and Low, High, or Max reasoning. Qwen3.8 Flash (qwen/qwen3.8-flash) is free on every tier for the first 24 hours after its August 27, 2026 MyTabulon launch, then requires Coding Plus+ at $0.15/M input, $0.016/M cached input, and $0.47/M output with a 1M-token context window, 131,072 max output, text/image/video input, and Low, Medium, or High reasoning. Paid Workspace Pro includes four weeks of Coding Plus and paid Workspace Max includes four weeks of Coding Pro for the payer's account only, never the whole team. Stable Maximo Pandora 3.8 Pro is available on Pro and Max at $2/M input, $0.50/M cached input, and $6/M output through 226K input tokens, then $5/M input, $0.80/M cached input, and $15/M output above 226K. Maximo Pandora 3.8 Preview starts on Plus and costs $1.50/M cache-miss input, $0.085/M cached input, and $4.20/M output, while Maximo Pandora 3.8 Nano is generally available on Plus, Pro, and Max. Z.ai GLM-5.3 (z-ai/glm-5.3) is available on Pro and Max at $1.40/M input, $0.26/M cached input, and $4.40/M output with a 1M-token context window, 128K max output, text-only input, and Low, High, or Max reasoning. The weekly usage pool is shared across every model available on the plan. Paid plans renew weekly. Model sets are strict supersets, so upgrades never remove access.
maximo-atlas-1.2maximo-atlas-1.3maximo-atlas-1.4maximo-pandora-3.8maximo-pandora-3.9maximo-pandora-3.8-nanomaximo-pandora-3.9-nanogoogle/gemini-3.5-flash-litemuse-spark-1.1muse-spark-1.2muse-spark-1.3muse-spark-1.2-contributordeepseek/deepseek-v4.1-flashdeepseek/deepseek-v4-flash-0731deepseek/deepseek-v4-flash-vision-expdeepseek/deepseek-v4-pro-0813openai/gpt-5.6-lunamaximo-pandora-3.8-prox-ai/grok-4.7x-ai/grok-4.6xiaomi/mimo-v2.6-flashxiaomi/mimo-v2.6-proz-ai/glm-5.3z-ai/glm-5.3-flashqwen/qwen3.8-flashgoogle/gemini-3.8-flashgoogle/gemini-3.7-flashmoonshotai/kimi-k3openai/gpt-5.6-terragoogle/gemini-3.1-proanthropic/claude-opus-5anthropic/claude-opus-5.5openai/gpt-5.6-solopenai/gpt-6-astraThe unified usage pool
Usage draws from one pool per workspace, with two limits. Your first request opens a fixed 5-hour window that fully resets when it ends; this is not a rolling average. A shared weekly pool resets every 7 days from the billing anchor and covers every model available on the plan. Usage is shown only as percentages. The selected model's published token pricing determines how quickly the pools move, so higher-priced models use them faster. The catalog shows cache-miss input, cached-input, and output rates for every model; when a provider does not publish a separate cache-read rate, cached tokens use the displayed input rate. Billing uses the provider-reported cached-token count and the active context or UTC pricing tier. GLM-5.3 Flash (z-ai/glm-5.3-flash) is free on every tier for the first 24 hours after its August 26, 2026 launch, then requires Coding Plus+ at $0.075/M input, $0.015/M cached input, and $0.25/M output. Qwen3.8 Flash (qwen/qwen3.8-flash) is free on every tier for the first 24 hours after its August 27, 2026 MyTabulon launch, then requires Coding Plus+ at $0.15/M input, $0.016/M cached input, and $0.47/M output. DeepSeek V4 Flash 0731 is $0.14/M cache-miss input, $0.0028/M cached input, and $0.28/M output until 16:00 UTC on August 16, 2026; from then, off-peak is $0.007/M cached input, $0.22/M cache-miss input, and $0.66/M output, while peak is $0.014/M cached input, $0.44/M cache-miss input, and $1.32/M output. DeepSeek V4 Pro 0813 is $0.003625/$0.435/$0.87 per million cached input/cache-miss input/output before the change, then $0.022/$0.66/$1.98 off-peak and $0.044/$1.32/$3.96 peak. Peak is 01:00–04:00 and 06:00–10:00 UTC; all other hours are off-peak. Maximo Atlas 1.2 costs $0.55/$0.05/$1.50 per million cache-miss input/cached input/output normally, with an automatic 80% sale at $0.11/$0.01/$0.30 from August 17 through August 31, 2026 UTC. Maximo Atlas 1.3 Preview costs $0.15/M input, $0.015/M cached input, and $0.45/M output standard, with an automatic launch sale at $0.11/$0.01/$0.30 per million through August 31, 2026 UTC. Stable Maximo Pandora 3.8 Pro costs $2/$0.50/$6 per million input/cached input/output through 226K input tokens, then $5/$0.80/$15 above 226K; the selected tier applies to the full request. Maximo Pandora 3.8 Preview costs $1.50/M cache-miss input, $0.085/M cached input, and $4.20/M output. Maximo Pandora 3.8 Nano is generally available at $0.15/M input, $0.015/M cached input, and $0.55/M output through 226K input tokens, then $0.25/M input, $0.025/M cached input, and $0.95/M output above 226K. The provider's full input count selects the Nano tier, and that tier applies to the full request. GLM-5.3 (z-ai/glm-5.3) is available on Pro and Max at $1.40/M input, $0.26/M cached input, and $4.40/M output with a 1M-token context window, 128K max output, text-only input, and Low, High, or Max reasoning. DeepSeek V4 Flash Vision (deepseek/deepseek-v4-flash-vision-exp) is available from Plus at $0.22/M input, $0.007/M cached input, and $0.66/M output with a 1M-token context window, 384K max output, and text and image input. Gemini 3.8 Flash (google/gemini-3.8-flash) is available on Pro and Max at $0.75/M input, $0.075/M cached input, and $3.75/M output — Google's introductory rates through December 31, 2026 — with a 1,048,576-token context window, 65,536 max output, text/image/video/audio input, and Low, Medium, or High reasoning; Gemini 3.7 Flash remains available at $0.375/M input, $0.0375/M cached input, and $1.875/M output. Meta Muse Spark 1.3 (muse-spark-1.3) is available from Coding Plus at $1.25/M input, $0.15/M cached input, and $4.25/M output with a 1,048,576-token context window, text/image/MP4-video/PDF input, and Minimal, Low, Medium, High, or XHigh reasoning; muse-spark-1.3-contributor routes the same model through Meta's direct endpoint and is not zero data retention. OpenAI GPT-6 Astra (openai/gpt-6-astra) is available on Coding Max at $10/M input, $1/M cached input, and $50/M output through 272K input tokens, then $20/M input, $2/M cached input, and $75/M output above 272K; the selected tier applies to the full request. It offers a 1,050,000-token context window, 128,000 max output, text/image/file input, and Low, Medium, High, XHigh, or Max reasoning. There is no Coding Plan context cap: each model keeps its full native context window.
Limit Resets
5-hour window Fully, 5h after your first request opens it
Weekly limit Every 7 days from your billing anchorWorkspace plan promotion
A paid Workspace Pro plan includes four weeks of Coding Plus, and a paid Workspace Max plan includes four weeks of Coding Pro. The promotion has a separate account-scoped pool and belongs only to the account that paid for the workspace plan; team members do not receive or consume it.
Developer referrals and Usage Boosts
Every workspace has a personal Coding Plan referral link. A referred developer receives 50% more five-hour and weekly usage during the first paid week. After that developer makes a successful Coding Plan API request, the referrer receives a claimable seven-day Usage Boost. A referrer on Coding Free receives a seven-day Coding Plus pass instead. Rewards must be claimed within 30 days, no more than two boosts can be active together, and no more than 10 activated referrals count in a rolling 30-day period. Self-referrals, refunded payments, disputes, and fraudulent activity are ineligible.
Share link → first valid paid week → invitee gets +50%
↓
first successful API request
↓
referrer claims +50% for 7 days
(or a 7-day Plus pass from Free)MyTabulon Developer Partners
Developers with an audience or community can apply from the Coding Plan or Referrals dashboard. Approved partners earn 25% of net Coding Plan subscription revenue for the first 12 successful weekly payments from each referred developer. Net revenue is the collected amount after discounts, taxes, processor fees, refunds, and chargebacks. Commission remains pending for 30 days and is then available for monthly payout. Nigerian NGN earnings use the existing referral withdrawal flow; international USD earnings use the partner payout request flow. Partner cash commission replaces the normal referrer Usage Boost, so a referral is never rewarded twice.
Quota headers & 429s
Every response carries x-codingplan-window-used-percent, x-codingplan-window-reset, x-codingplan-weekly-used-percent, and x-codingplan-weekly-reset. A depleted pool returns HTTP 429 with rate_limit_exceeded and Retry-After. Reaching your tier's parallel-request allowance returns concurrency_limit_exceeded. No internal unit or credit balance is exposed.
Data retention and ZDR
A small number of Coding Plan models are not zero data retention (no ZDR). For these models your prompts, tool calls, and outputs are logged and retained by the model's provider and by Maximo AI for safety, abuse prevention, and product improvement. In the dashboard model table the ID shows a **No ZDR** pill — tap it for the full per-model notice. Other Coding Plan models keep their provider's retention policy; ZDR remains available where the upstream provider offers it. Avoid sending secrets or regulated data to any non-ZDR model.
Model value guide and benchmark policy
The dashboard and pricing page order models by Coding Plan value using current public token pricing, native context, tool and multimodal support, and available published coding or agent benchmarks. It is a buying guide, not a synthetic intelligence score. Every displayed score keeps its benchmark name, source, and date. Missing comparable official scores are shown as not published, and preview claims are never converted into benchmark numbers.
/chat/completionsai.codingOpenAI-compatible Chat Completions for every model included in the workspace's Coding Plan tier. Supports streaming, function tools, images, files, and structured output without injected prompts or tools. A few models are not zero data retention — a No ZDR pill appears next to their model ID, and tapping it shows the full per-model retention notice.
{
"model": "maximo-atlas-1.2",
"messages": [
{ "role": "user", "content": [
{ "type": "text", "text": "What does this diagram mean?" },
{ "type": "image_url", "image_url": { "url": "data:image/png;base64,..." } }
]}
],
"tools": [ { "type": "function", "function": { "name": "run_tests", "parameters": { "type": "object", "properties": {} } } } ],
"stream": true
}data: {"id":"chatcmpl-...","object":"chat.completion.chunk","model":"maximo-atlas-1.2","choices":[{"index":0,"delta":{"content":"The"},"finish_reason":null}]}
data: {"id":"chatcmpl-...","object":"chat.completion.chunk","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"id":"call_...","function":{"name":"run_tests","arguments":"{}"}}]},"finish_reason":null}]}
data: {"id":"chatcmpl-...","object":"chat.completion.chunk","choices":[],"usage":{"prompt_tokens":412,"completion_tokens":128,"total_tokens":540}}
data: [DONE]/responsesai.codingOpenAI-compatible Responses API for Codex and other Responses clients. Accepts input items, instructions, custom function tools, tool results, reasoning effort, structured text formats, and typed SSE streaming events. Stored-response continuation and background mode are not supported; send conversation items in input with store: false.
{
"model": "maximo-atlas-1.2",
"input": "Inspect this repository and fix the failing test.",
"tools": [{
"type": "function",
"name": "shell",
"description": "Run a shell command",
"parameters": {
"type": "object",
"properties": { "command": { "type": "string" } },
"required": ["command"]
}
}],
"stream": true,
"store": false
}event: response.created
data: {"type":"response.created","response":{"id":"resp_...","status":"in_progress"}}
event: response.output_text.delta
data: {"type":"response.output_text.delta","delta":"I’ll inspect the tests."}
event: response.completed
data: {"type":"response.completed","response":{"id":"resp_...","status":"completed","usage":{"input_tokens":420,"output_tokens":86}}}/modelsany valid keyOpenAI-format list of models available to the current Coding Plan tier, including native context, capabilities, reasoning efforts, preview state, minimum tier, recommendation, and public token pricing.
{
"object": "list",
"data": [{
"id": "maximo-atlas-1.2",
"object": "model",
"owned_by": "mytabulon",
"context_window": 1000000,
"capabilities": ["text", "images", "files", "tools", "streaming"],
"preview": false,
"minimum_coding_plan": "free"
}],
"coding_plan": { "tier": "free", "concurrency": 2 }
}/coding-plan/usageany valid keyLive quota for both limits — used and remaining percentage, window start, and both reset timestamps — plus a 14-day daily usage series.
{
"object": "coding_plan.usage",
"active": true,
"plan_id": "coding_plus_v1",
"tier": "plus",
"models": ["maximo-atlas-1.2", "maximo-pandora-3.8", "maximo-pandora-3.8-nano"],
"concurrency": 10,
"referralBoostPercent": 50,
"referralBoostEndsAt": "2026-07-17T09:00:00.000Z",
"quota_type": "unified_pool",
"reset_window": "5h_fixed",
"window": { "usedPercent": 12.4, "remainingPercent": 87.6, "startedAt": "2026-07-10T13:00:00.000Z", "resetAt": "2026-07-10T18:00:00.000Z" },
"weekly": { "usedPercent": 61.2, "remainingPercent": 38.8, "startedAt": "2026-07-07T09:00:00.000Z", "resetAt": "2026-07-14T09:00:00.000Z" },
"daily": [ { "day": "2026-07-09", "requests": 42, "used_percent": 4.6 } ]
}Referral management API
Referral management uses the authenticated MyTabulon application API rather than an mtb_live_ model key. The owner of the workspace can bind a referral before the first paid Coding Plan, claim an available reward, apply for the Developer Partner program, and request an international partner payout. The Coding Plan status response includes the same referral overview used by the dashboard.
GET /subscriptions/companies/{companyId}/coding-plan/referrals
POST /subscriptions/companies/{companyId}/coding-plan/referrals/bind
POST /subscriptions/companies/{companyId}/coding-plan/referrals/rewards/{rewardId}/claim
POST /subscriptions/companies/{companyId}/coding-plan/developer-partner/apply
POST /subscriptions/companies/{companyId}/coding-plan/developer-partner/payoutsChoose your integration
For the fastest start, use the Maximo Syntax CLI — install it, sign in with MyTabulon, and your Coding Plan links automatically. Or use the dedicated Codex, Kilo Code, and OpenClaw pages for copy-paste configuration and verification steps. For any other OpenAI-compatible client, set the base URL to https://api.mytabulon.com/v1, provide a live key with ai.coding, and select a model returned by GET /models. MyTabulon does not append a system prompt or inject tools.
curl https://api.mytabulon.com/v1/chat/completions \
-H "Authorization: Bearer mtb_live_..." \
-H "Content-Type: application/json" \
-d '{
"model": "maximo-atlas-1.2",
"messages": [{"role": "user", "content": "Write a binary search in TypeScript"}]
}'