{"slug": "adieu-claude", "title": "Adieu, Claude", "summary": "Developer Simon Willison replaced Anthropic's Claude Code with OpenCode after losing trust in Anthropic due to the Fable 5 silent downgrade controversy and export-control issues. He cites open standards and open source as safeguards against vendor lock-in, noting OpenCode's MIT license and compatibility with existing Claude Code skills.", "body_md": "# Adieu, Claude\n\nWhen Anthropic proved to be a liability, I replaced Claude Code with OpenCode.\n\nI already ran [Crossfire](/crossfire/) and a few private projects through OpenRouter on a pay-as-you-go basis.\nClaude Code was the subscription I did not have to think about for twenty bucks a month, but I no longer trust the company behind it.\n\n##\nFable’s fairy tale\n\nAnthropic shipped Fable 5, silently rerouting certain requests to the weaker Opus 4.8.\nThe affected requests were related to cybersecurity and anything resembling model distillation.\nAnthropic admitted [it had made the wrong decision](https://fortune.com/2026/06/11/anthropic-fable-5-silent-downgrade-backlash-national-security-transparency/) once the backlash made it impossible not to.\n\nA few days later, the export-control directive grounded Fable 5, for reasons Anthropic itself called a misunderstanding. I do not much care whose fault that part is. Governments issue directives.\n\n##\nA fairy tale retold\n\nI had already dropped OpenAI before this.\nThey claimed GPT-2 was too dangerous to release, which was rubbish, and the “Open” was gone the moment they smelled stacks of money.\nThen [they put ads in ChatGPT](https://openai.com/index/testing-ads-in-chatgpt/).\nWhen Anthropic refused the Pentagon’s demand to remove red lines on fully autonomous weapons, OpenAI stepped in hours later and [signed the deal](https://www.npr.org/2026/02/27/nx-s1-5729118/trump-anthropic-pentagon-openai-ai-weapons-ban).\n\nTwo weeks after Anthropic’s models were pulled, OpenAI announced its latest models and then restricted them to “trusted partners” [at the US government’s request](https://www.cnbc.com/2026/06/26/openai-limits-new-ai-models-to-trusted-partners-request-us-government.html).\nThe US has made it impossible for me to treat access to frontier models as a sure thing.\n\n##\nCrossing the drawbridge\n\nThe only reason the switch from Claude Code to OpenCode was painless is that I built on open foundations before I needed them, which is why [migrating off Logseq](/goodbye-logseq/) was relatively painless, too.\nThe coding agent talks to Zed through the [Agent Client Protocol](https://opencode.ai/docs/acp/), an open standard.\nI had built more than a dozen skills for Claude Code, yet OpenCode read them without a single change, because they are plain Markdown with YAML front matter.\nOpenCode is MIT-licensed and I could run it myself if I wanted to.\nNone of that was true with Anthropic or OpenAI.\n\nLock-in is the moat, not the data or the model architecture. It is to keep customers from leaving, not necessarily to keep competitors from entering. That moat is what keeps you captive when the company decides, as it inevitably does, to squeeze customers. Open standards and open source are the only guarantee against that betrayal.\n\n##\nThe toll\n\nI already knew what pay-as-you-go felt like.\nAt the end of the month I often asked myself whether it was worth it.\nClaude Code had a flat fee and I never asked that question.\n[OpenCode Go](https://opencode.ai/go?ref=PET2T3Y3AD) ($5 referral code) gives me the same at ten dollars a month, and they are honest about what that buys: “generous limits and reliable access to the most capable open source models.”\nTwenty dollars a month for OpenAI or Anthropic is twice the price, and the $100–$200 plans are in a different league entirely.\nYou get what you pay for: a single prompt against GLM-5.2 on a large codebase burns through the quota like an AI startup through $100m, which is where the truly free models fill the gap.\n\nThe models are good enough for the work I described.\nDay to day, I use Big Pickle, a model OpenCode will not name and only calls [a stealth model](https://opencode.ai/docs/zen/), free for now while the team collects feedback.\nIt handles basic refactors, scripts, and multi-file edits without a fuss.\nFor harder tasks I switch to GLM-5.2, which [scored 81 against Claude Opus 4.8’s 85 and GPT-5.5’s 84](https://huggingface.co/blog/zai-org/glm-52-blog) ahead of Gemini 3.1 Pro (74), though of course [benchmarks are not everything](/the-ai-benchmark-trap/).\n\nI could buy the same GLM models through [Z.ai](https://docs.z.ai/devpack/overview), but its platform [separates Chinese and overseas infrastructure](https://github.com/THUDM/z-ai-sdk-python) and reliability complaints are [rife](https://www.trustpilot.com/review/z.ai).\nOpenCode Go runs the same models through servers in the US, EU, and Singapore instead.\nThat distinction is the entire reason I am comfortable using GLM at all.\n\n##\nThe moral of the story\n\nNone of this makes me independent of frontier labs in any complete sense.\nThe harder tasks still send me back to Zen, OpenCode’s pay-as-you-go option, where Claude and GPT are available—Gemini is there too, but its tendency to roll over like a puppy makes it useless when I need it to disagree with me.\nThe only full exit from that arrangement is [running open models on my own hardware](/local-llms-for-zed-and-obsidian/), which is fine for autocomplete and short questions though not yet usable for anything that requires the model to “think.”\nThe OpenCode quotas are a bit of a bummer, but money can fix that problem.\nTrust cannot be bought, and that is the moral of the story.", "url": "https://wpnews.pro/news/adieu-claude", "canonical_source": "https://ianreppel.org/adieu-claude/", "published_at": "2026-06-27 22:00:00+00:00", "updated_at": "2026-07-13 13:52:38.283301+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-ethics", "ai-policy", "developer-tools", "ai-products"], "entities": ["Anthropic", "OpenCode", "Claude Code", "OpenAI", "OpenRouter", "Zed", "GLM-5.2", "Big Pickle"], "alternates": {"html": "https://wpnews.pro/news/adieu-claude", "markdown": "https://wpnews.pro/news/adieu-claude.md", "text": "https://wpnews.pro/news/adieu-claude.txt", "jsonld": "https://wpnews.pro/news/adieu-claude.jsonld"}}