Run cheap DeepSeek Flash workers inside Claude Code with Opus as the boss
A builder plugged DeepSeek Flash into Claude Code as the worker model and kept Opus 5.5 as the lead that plans and checks. He cites DeepSeek's own chart where Flash scored 77.0 vs 74.0 on the same agent benchmark.
Source
@zer0point_eth on X · 2026-10-19
Written in our own words. Credit to the original author. Link checked 2026-10-22.
What they did
A builder plugged DeepSeek Flash into Claude Code as the worker model and kept Opus 5.5 as the lead that plans and checks. He cites DeepSeek's own chart where Flash scored 77.0 vs 74.0 on the same agent benchmark.
Result (as claimed)
The same agent task costs about $0.20 on Flash versus about $10 on Opus, by his numbers.
Tools they used
Cost: Claude Code: from $17/mo; Claude Opus 5.5: from $4 in / $20 out per 1M tokens.
Set it up
Pick a contained repo or task
Start on a branch or side project where mistakes are cheap.
Install and configure
Set up Claude Code, DeepSeek, Claude Opus 5.5. Read what it can run on your machine and limit permissions where possible.
Run it on a real task
Give it a well-scoped issue and let it work. Review the diff like any pull request.
Add guardrails
Keep tests and CI in the loop, and never let an agent push to main or touch production secrets unattended.
Copy the full setup prompt
Paste into your agent and it will walk you through the setup, asking before anything that sends, posts or spends.
You are helping me set up this AI use case end to end in Claude Code. Goal: Run cheap DeepSeek Flash workers inside Claude Code with Opus as the boss What the original author did: A builder plugged DeepSeek Flash into Claude Code as the worker model and kept Opus 5.5 as the lead that plans and checks. He cites DeepSeek's own chart where Flash scored 77.0 vs 74.0 on the same agent benchmark. Reported result (as claimed by the author): The same agent task costs about $0.20 on Flash versus about $10 on Opus, by his numbers. Source: https://x.com/zer0point_eth/status/2112219791381479492 Tools they used: Claude Code, DeepSeek, Claude Opus 5.5 Category: Dev & ops Please: 1. Ask me 3 short questions about my situation (accounts I have, budget, what I want the output to look like). 2. Propose a step-by-step plan using these tools or cheaper alternatives, with the time and cost of each step. Use official pricing pages; say "check pricing" if unsure. 3. Then do the setup with me one step at a time, waiting for my OK before anything that sends messages, posts publicly, spends money or deletes data. 4. Finish with a checklist to run it again next time.
You are helping me set up this AI use case end to end in Codex. Goal: Run cheap DeepSeek Flash workers inside Claude Code with Opus as the boss What the original author did: A builder plugged DeepSeek Flash into Claude Code as the worker model and kept Opus 5.5 as the lead that plans and checks. He cites DeepSeek's own chart where Flash scored 77.0 vs 74.0 on the same agent benchmark. Reported result (as claimed by the author): The same agent task costs about $0.20 on Flash versus about $10 on Opus, by his numbers. Source: https://x.com/zer0point_eth/status/2112219791381479492 Tools they used: Claude Code, DeepSeek, Claude Opus 5.5 Category: Dev & ops Please: 1. Ask me 3 short questions about my situation (accounts I have, budget, what I want the output to look like). 2. Propose a step-by-step plan using these tools or cheaper alternatives, with the time and cost of each step. Use official pricing pages; say "check pricing" if unsure. 3. Then do the setup with me one step at a time, waiting for my OK before anything that sends messages, posts publicly, spends money or deletes data. 4. Finish with a checklist to run it again next time.
You are helping me set up this AI use case end to end in Grok Bot. Goal: Run cheap DeepSeek Flash workers inside Claude Code with Opus as the boss What the original author did: A builder plugged DeepSeek Flash into Claude Code as the worker model and kept Opus 5.5 as the lead that plans and checks. He cites DeepSeek's own chart where Flash scored 77.0 vs 74.0 on the same agent benchmark. Reported result (as claimed by the author): The same agent task costs about $0.20 on Flash versus about $10 on Opus, by his numbers. Source: https://x.com/zer0point_eth/status/2112219791381479492 Tools they used: Claude Code, DeepSeek, Claude Opus 5.5 Category: Dev & ops Please: 1. Ask me 3 short questions about my situation (accounts I have, budget, what I want the output to look like). 2. Propose a step-by-step plan using these tools or cheaper alternatives, with the time and cost of each step. Use official pricing pages; say "check pricing" if unsure. 3. Then do the setup with me one step at a time, waiting for my OK before anything that sends messages, posts publicly, spends money or deletes data. 4. Finish with a checklist to run it again next time.
The Daily 5
Get five new use cases tomorrow morning
Free. Real sources. Unsubscribe any time.
Guides for this use case
How to vibe code an app with Claude Code · How to compare AI models on the same prompt
Related use cases
Make your smart TV dumb again with an AI agent
TechCrunch rounded up the trend: people point Claude Code or another agent at open-source repos like dumbtv and cleanyourtv to remove sponsored rows, ACR viewing trackers and bloat from Android TV, Google TV and Fire TV sets, often swapping in a cleaner launcher.
Turn a spare Android TV box into an always-on host for your AI agent
A builder had Claude Code inspect the Android TV box from his TV provider from a prompt and one photo, install a Linux shell built for its chip, add a wake lock and battery whitelist so Android cannot kill the agent, and wire the agent to his own Telegram bot with no open ports.
Run eight AI coding CLIs side by side in persistent rooms on your own machine
SpaceApp is a self-hosted workspace that runs AI coding CLIs side by side in persistent rooms: OpenCode, Codex, Claude Code, Gemini, Qwen, Kimi, Grok and DeepSeek, each in its own pane on your own machine.
Measure how much of your usage limit resuming a long session costs
A developer tested resuming long sessions in Claude Code with Opus 5.5 and in Codex with GPT-6.1 Sol, both on medium effort, and watched the usage meters. Resuming a long Claude Code session took a big chunk of the 5-hour limit, while codex resume did not show the same jump.