Running Qwen3 Coder 30B as a cheap subagent for Claude Code, Codex and opencode

A developer demonstrated running Qwen3 Coder 30B, a 30B mixture-of-experts coding model with tool calling, as a low-cost subagent behind Claude Code, Codex CLI and opencode via an OpenAI-compatible endpoint on Synap. In a Codex test, fixing a one-character bug in is_even() consumed 49,245 input tokens and 722 output tokens, costing roughly $0.0035 per task at $0.0675 per 1M input and $0.27 per 1M output, with the developer noting that agent system prompts and tool definitions dominate token use. The writeup also documents a Claude Code run that fixed a bug in six turns and 11.3 seconds, and a free tier of 100,000 tokens and 50 requests per day.

Big frontier models are great at planning a change across a codebase. They are expensive overkill for "add type hints to this file", "fix the off-by-one in is even " or "write two doctests". This post shows how to hand that second kind of work to Qwen3 Coder 30B a 30B mixture-of-experts coding model with tool calling through an OpenAI-compatible endpoint. It covers three agents, Claude Code, Codex CLI and opencode, plus a 40-line Python "subagent" you can call from any of them. Everything below was run on 2026-11-11 against https://synap402.com/v1 Synap https://synap402.com , the API we build . The token counts and prices are copied from those runs. From curl -s https://synap402.com/v1/models today: | model | input / 1M tokens | output / 1M tokens | batch 24h in / out | context | |---|---|---|---|---| | qwen3-coder-30b | $0.0675 | $0.27 | $0.03375 / $0.135 | 98,304 | It supports tools , tool choice auto , required , response format with json schema , and streaming. There is also a free tier on this model: 100,000 tokens and 50 requests a day per account Google sign-in, no card . It allows 10 requests a minute per key and 4,096 output tokens per answer. As you'll see below, one coding-agent task can use half of that, so it's for trying things out, not daily work. Claude Code speaks Anthropic's Messages API, and the endpoint serves /v1/messages too. Five environment variables: export ANTHROPIC BASE URL=https://synap402.com no /v1, Claude Code appends /v1/messages export ANTHROPIC AUTH TOKEN=$SYNAP API KEY export ANTHROPIC MODEL=qwen3-coder-30b export ANTHROPIC DEFAULT HAIKU MODEL=qwen3-coder-30b its own background tasks export CLAUDE CODE MAX CONTEXT TOKENS=98304 otherwise it warns about an unknown model I gave it a file with a deliberate bug: python mathx.py def add a, b : return a - b claude -p "mathx.py has a bug in add . Fix it, then reply with one line saying what you changed." \ --allowedTools "Read,Edit" --output-format json Result on Claude Code 2.1.322: 6 turns, 11.3 seconds, "is error": false , and the file now says return a + b . The model read the file, called Claude Code's own Edit tool and reported back. Claude Code also prints a costUSD estimate for a model it doesn't know. Ignore it and use the usage the API reports instead. Codex talks to the Responses API /v1/responses . Add one provider block to ~/.codex/config.toml : model provider = "synap" model = "qwen3-coder-30b" model providers.synap name = "Synap" base url = "https://synap402.com/v1" env key = "SYNAP API KEY" wire api = "responses" Then: python printf 'def is even n :\n return n %% 2 == 1\n' evens.py codex exec --json --skip-git-repo-check -s workspace-write \ "evens.py has a bug in is even . Fix it." </dev/null | grep turn.completed On codex-cli 0.166.0 that printed: {"type":"turn.completed","usage":{"input tokens":49245,"cached input tokens":0,"output tokens":722,"reasoning output tokens":0}} and evens.py now returns n % 2 == 0 . Without </dev/null , codex exec sits waiting on stdin. Look at that input number: 49,245 tokens to fix one character . Almost all of it is Codex's system prompt and tool definitions, not your code. At $0.0675 per 1M input and $0.27 per 1M output: 49,245 × $0.0675 / 1M = $0.003324 722 × $0.27 / 1M = $0.000195 total ≈ $0.0035 per task So 200 small tasks like this a day comes to about $0.70. The same overhead is why one Codex task uses half the free daily allowance. Agent prompts are large, so per-input-token price matters much more than it does in chat. opencode doesn't read /v1/models for a custom provider, so you list the model ids yourself. Put this in opencode.json in the project root. {env:...} keeps the key out of the file: { "$schema": "https://opencode.ai/config.json", "provider": { "synap": { "npm": "@ai-sdk/openai-compatible", "name": "Synap", "options": { "baseURL": "https://synap402.com/v1", "apiKey": "{env:SYNAP API KEY}" }, "models": { "qwen3-coder-30b": { "name": "Qwen3 Coder 30B" } } } }, "model": "synap/qwen3-coder-30b" } python printf 'def square x :\n return x 2\n' sq.py npx -y opencode-ai@latest run "sq.py: square is wrong. Fix it." </dev/null On opencode 1.19.5 it changed the function to x x , then ran its own check over eight inputs square -3 = 9 … before it answered. The first run is slow because opencode downloads the provider package then. Running a whole session on a 30B model is fine for routine work. The more interesting setup is to keep your strong model as the planner and have it hand scoped edits to the small one. Every agent above can run a shell command, so the subagent can be a plain script: delegate.py: hand one scoped coding task to a small model, get JSON back, print the cost import json, os, sys, urllib.request from openai import OpenAI BASE = "https://synap402.com/v1" MODEL = "qwen3-coder-30b" client = OpenAI base url=BASE, api key=os.environ "SYNAP API KEY" def price model id : """Live USD per 1M tokens input, output from the public catalog.""" with urllib.request.urlopen f"{BASE}/models" as r: models = json.load r "data" p = next m for m in models if m "id" == model id "pricing" return p "prompt per million" , p "completion per million" def delegate task, path : source = open path .read resp = client.chat.completions.create model=MODEL, temperature=0.2, messages= {"role": "system", "content": "You are a careful Python engineer. Change only what the task asks."}, {"role": "user", "content": f"Task: {task}\n\nFile {path}:\n python\n{source}\n "}, , response format={ "type": "json schema", "json schema": { "name": "edit", "schema": { "type": "object", "properties": { "summary": {"type": "string"}, "new file": {"type": "string"}, }, "required": "summary", "new file" , }, }, }, out = json.loads resp.choices 0 .message.content p in, p out = price MODEL u = resp.usage cost = u.prompt tokens p in / 1e6 + u.completion tokens p out / 1e6 return out, u, cost if name == " main ": out, u, cost = delegate sys.argv 1 , sys.argv 2 with open sys.argv 2 , "w" as f: f.write out "new file" print out "summary" print f"{u.prompt tokens} in / {u.completion tokens} out - ${cost:.6f}" Input file: python import re def slugify title : s = title.lower s = re.sub r" ^a-z0-9 +", "-", s return s pip install openai python delegate.py "Strip leading/trailing dashes, add type hints and a docstring with two doctest examples." slugify.py python -m doctest -v slugify.py Output: Added type hints, docstring with doctest examples, and modified the function to strip leading/trailing dashes as requested. 93 in / 202 out - $0.000061 ... 2 passed and 0 failed. The edit runs without an agent harness, so there's no 49k-token system prompt: 93 tokens in, 202 out, six thousandths of a cent. The structured output means the caller gets a file it can write, not prose it has to parse. And the doctests the model wrote pass. To wire it in, put one line in CLAUDE.md / AGENTS.md : "For single-file mechanical edits type hints, docstrings, renames, small bug fixes, test scaffolds , run python delegate.py "<task " <file and review the diff instead of editing yourself." From these runs and day-to-day use: Paid keys start at $10, and top-ups don't expire. You can also pay in USDC through x402 with no account. Prompts and completions are not stored. Synap by Linkrra: one OpenAI-compatible API for open models. synap402.com https://synap402.com