Claude Limits Got Brutal. Here's Where to Take Your $200

Claude usage limits dropped again on September 14, and the $200 Max plan now buys less than it did last month. Here is where that $200 goes further for AI coding right now. I price out every real option: the move to ChatGPT Pro and Codex, splitting $100 of Claude Max with $100 of ChatGPT Pro, one frontier plan plus a cheap open-weight workhorse like GLM-5.3 Flash or DeepSeek V4.1 Flash in OpenCode, and running Qwen locally. Then the two dark horses, Grok and Gemini, the token housekeeping that stretches any plan, and where I would put the money this week. 📬 Get the hotter takes in your inbox every Tuesday: https://devsplainers.com/takeouts/ ⏱️ Chapters 00:00 Claude limits after September 14 00:39 What $200 buys at Anthropic now 01:08 The ChatGPT and Codex sidestep 02:06 Option 1: split $100 Claude + $100 ChatGPT 03:05 Option 2: frontier plan + cheap workhorse 04:36 Option 3: run it locally 05:15 Dark horse: Grok 06:04 Dark horse: Gemini and Antigravity 06:54 Save tokens on any plan 07:36 Where I'd put the money 📺 Related videos Your AI Coding Plan Is Now Worth Less: https://youtu.be/thXxgXLfW6Y OpenCode vs Claude Code and Codex: The Harness War: https://youtu.be/KpU6Zmdl9g4 How DeepSeek V4.1 Flash Killed Its Own Flagship: https://youtu.be/Weom9fQnnJ0 Local LLM Coding: Can a Base MacBook Handle It?: https://youtu.be/jr1qTGOxc1k What happened to Claude usage limits? Anthropic's temporary +50% Claude Code promotion ended on September 13, 2026. From September 14 the weekly limits settled at a permanent +25% over the old baseline, which Anthropic itself says works out to a 17% reduction from the promo level. Max 20x users report the plan feeling like the old 5x, and Max 5x users report running out of weekly usage by day 2. OpenAI paused new sign-ups for its $200 ChatGPT Pro plan in the same week, so the usual Claude alternative is harder to reach too. In this video: • Claude Max vs ChatGPT Pro: the $100 + $100 split • ChatGPT Pro and Codex limits with GPT-6 Astra • Cheap coding models: GLM-5.3 Flash and DeepSeek V4.1 Flash • Local coding models: Qwen 3.8 27B and the RTX 5090 math • Grok 4.6, Gemini and Antigravity as dark horses • Claude Code token tips that work on any plan 📚 Sources Anthropic 17% weekly limit reduction: https://www.bleepingcomputer.com/news/artificial-intelligence/anthropic-is-cutting-claude-codes-current-weekly-limits-by-17-percent/ Claude Code weekly limits promotion: https://support.claude.com/en/articles/15910845-claude-code-may-august-2026-weekly-limits-promotion Claude pricing: https://claude.com/pricing Max 5x vs 20x pool measurement: https://www.reddit.com/r/ClaudeAI/comments/1vx0k69/lifting_the_curtain_the_max_x5_and_max_x20_usage/ Post-Sep-14 user reports: https://www.reddit.com/r/ClaudeCode/comments/1wfwl6k/the_limits_have_been_reduced_even_further_now_its/ Codex pricing: https://learn.chatgpt.com/docs/pricing Astra quota burn from Codex logs: https://www.reddit.com/r/codex/comments/1wgs7yo/your_codex_logs_track_your_quota_on_every_turn_i/ OpenAI pauses $200 Pro sign-ups: https://fortune.com/2026/09/11/openai-astra-chatgpt-pro-pause/ Claude subscriptions blocked in third-party tools: https://alternativeto.net/news/2026/4/anthropic-ends-free-claude-access-for-third-party-tools-like-openclaw Z.ai pricing: https://docs.z.ai/guides/overview/pricing GLM-5.3 Flash as Ox Alpha: https://x.com/jietang/status/2092850258573471936 OpenRouter spend report: https://news.ycombinator.com/item?id=49351170 DeepSeek cost per task across harnesses: https://www.reddit.com/r/PiCodingAgent/comments/1vqs0c7/ Local payback estimate: https://news.ycombinator.com/item?id=49706656 Terminal-Bench scores: https://www.deeplearning.ai/the-batch/groks-cursor-alliance-pays-off xAI model pricing: https://docs.x.ai/docs/models SuperGrok Heavy vs Claude Max token week: https://www.reddit.com/r/GrokBuild/comments/1vwr22r/ Grok Build CLI upload incident: https://cybersecuritynews.com/xai-grok-build-cloud-storage/ Google AI plans: https://one.google.com/about/google-ai-plans/ Antigravity third-party harness bans: https://news.ycombinator.com/item?id=49548452 Google engineers get Claude in Antigravity: https://www.techrepublic.com/article/news-google-claude-internal-coding-gemini/ Fable sub-agent spawn report: https://www.reddit.com/r/ClaudeCode/comments/1wh442e/an_hour_and_a_half_into_20x_plan_one_fable_51/ Claude Code costs, cache and /usage: https://code.claude.com/docs/en/costs Matt Pocock on trimming the system prompt: https://www.youtube.com/watch?v=oLx4yCbeklQ AGENTS.md adoption: https://www.reddit.com/r/AgentsOfAI/comments/1vylrj6/ #ClaudeCode #Codex #AICoding #ChatGPT #DevTools