Broken Claude Max

Description

As of a few moments ago, usage of claude max stopped with the following error:

I did try to reconnect, but got the same error.

Plugins

No response

OpenCode version

1.1.8

Steps to reproduce

No response

Screenshot and/or share link

No response

Operating System

mac

Terminal

No response

This issue might be a duplicate of existing issues. Please check:

Claude Max not working #4660: Claude Max not working - user reports Claude Max stopped responding, similar to this issue's reconnection failure
Claude subscription token expires after a period of time #6559: Claude subscription token expires after a period of time - could be an underlying cause of connection failures
Issue: OAuth + claude-opus-4-5 returns HTTP 500 #6637: OAuth + claude-opus-4-5 returns HTTP 500 - related authentication/model usage failure pattern

Feel free to ignore if none of these address your specific case.

Me too. It was working fine 1 hour ago and now I am getting this error. I am on version 1.0.220. Claude status is up, but not sure if its a provider issue. Gpt 5.2 with the oauth plugin works for me

Having the same issue, been using it all afternoon/evening and just got this.

me too, working fine yesterday

Same. Downgraded my claude plan to $100 immediately because I won't be able to use it as much without opencode.

Honestly using CC is like going back to stone age. Gonna be rough until if/when this gets fixed.

EDIT: I've cancelled my Anthropic subscription completely. Trying to use CC for the last 10 minutes and it's unusuable for the workflows I have and the visibility I need that opencode has enabled me to have. Will explore other models.

Same

got a message too

came here to find exactly this

Same here... Can't believe anthropic would shoot themselves in the foot like this...

@arthur404dev

@arthur404dev technically, this has nothing to do with anthropic. I don't believe we are supposed to be able to use the max plan outside of anthropic tools the way we are.

same ；；

it's likely they blocked it - i don't have time right now to investigate how but it's possible they blocked opencode specifically

I canceled my Claude Max subscription, I am using Codex now.

✅ Workaround confirmed working

Added mode override in opencode.json to bypass the pattern matching:

"mode": {
"build": {
"prompt": "You are Claude Code, Anthropic's official CLI for Claude."
},
"plan": {
"prompt": "You are Claude Code, Anthropic's official CLI for Claude."
}
}

Tested with: Anthropic Max Subscription + claude-opus-4-5 and claude-sonnet-4-5

This replaces the system prompt identifier that Anthropic is blocking. Works as of 2026-01-28.

Copy-paste ready: See commit with full opencode.json

Reference: anomalyco/opencode-anthropic-auth#17 (comment)

? in the .config directory opencode.json I do not see where to add or put

{
  "plugin": [
    "oh-my-opencode",
    "opencode-antigravity-auth@1.3.2",
    "opencode-ultraplan",
    "opencode-openai-codex-auth",
    "opencode-anthropic-auth@0.0.8",
    "opencode-claude-cli"
  ],
  "provider": {
    "google": {
      "name": "Google",
      "models": {
        "antigravity-gemini-3-pro-high": {
          "name": "Gemini 3 Pro High (Antigravity)",
          "thinking": true,
          "attachment": true,
          "limit": {
            "context": 1048576,
            "output": 65535
          },
          "modalities": {
            "input": [
              "text",
              "image",
              "pdf"
            ],
            "output": [
              "text"
            ]
          }
        },
        "antigravity-gemini-3-pro-low": {
          "name": "Gemini 3 Pro Low (Antigravity)",
          "thinking": true,
          "attachment": true,
          "limit": {
            "context": 1048576,
            "output": 65535
          },
          "modalities": {
            "input": [
              "text",
              "image",
              "pdf"
            ],
            "output": [
              "text"
            ]
          }
        },
        "antigravity-gemini-3-flash": {
          "name": "Gemini 3 Flash (Antigravity)",
          "attachment": true,
          "limit": {
            "context": 1048576,
            "output": 65536
          },
          "modalities": {
            "input": [
              "text",
              "image",
              "pdf"
            ],
            "output": [
              "text"
            ]
          }
        }
      }
    },
    "openai": {
      "name": "OpenAI",
      "options": {
        "reasoningEffort": "medium",
        "reasoningSummary": "auto",
        "textVerbosity": "medium",
        "include": [
          "reasoning.encrypted_content"
        ],
        "store": false
      },
      "models": {
        "gpt-5.2": {
          "name": "GPT 5.2 (OAuth)",
          "limit": {
            "context": 272000,
            "output": 128000
          },
          "modalities": {
            "input": [
              "text",
              "image"
            ],
            "output": [
              "text"
            ]
          },
          "variants": {
            "none": {
              "reasoningEffort": "none",
              "reasoningSummary": "auto",
              "textVerbosity": "medium"
            },
            "low": {
              "reasoningEffort": "low",
              "reasoningSummary": "auto",
              "textVerbosity": "medium"
            },
            "medium": {
              "reasoningEffort": "medium",
              "reasoningSummary": "auto",
              "textVerbosity": "medium"
            },
            "high": {
              "reasoningEffort": "high",
              "reasoningSummary": "detailed",
              "textVerbosity": "medium"
            },
            "xhigh": {
              "reasoningEffort": "xhigh",
              "reasoningSummary": "detailed",
              "textVerbosity": "medium"
            }
          }
        },
        "gpt-5.2-codex": {
          "name": "GPT 5.2 Codex (OAuth)",
          "limit": {
            "context": 272000,
            "output": 128000
          },
          "modalities": {
            "input": [
              "text",
              "image"
            ],
            "output": [
              "text"
            ]
          },
          "variants": {
            "low": {
              "reasoningEffort": "low",
              "reasoningSummary": "auto",
              "textVerbosity": "medium"
            },
            "medium": {
              "reasoningEffort": "medium",
              "reasoningSummary": "auto",
              "textVerbosity": "medium"
            },
            "high": {
              "reasoningEffort": "high",
              "reasoningSummary": "detailed",
              "textVerbosity": "medium"
            },
            "xhigh": {
              "reasoningEffort": "xhigh",
              "reasoningSummary": "detailed",
              "textVerbosity": "medium"
            }
          }
        },
        "gpt-5.1-codex-max": {
          "name": "GPT 5.1 Codex Max (OAuth)",
          "limit": {
            "context": 272000,
            "output": 128000
          },
          "modalities": {
            "input": [
              "text",
              "image"
            ],
            "output": [
              "text"
            ]
          },
          "variants": {
            "low": {
              "reasoningEffort": "low",
              "reasoningSummary": "detailed",
              "textVerbosity": "medium"
            },
            "medium": {
              "reasoningEffort": "medium",
              "reasoningSummary": "detailed",
              "textVerbosity": "medium"
            },
            "high": {
              "reasoningEffort": "high",
              "reasoningSummary": "detailed",
              "textVerbosity": "medium"
            },
            "xhigh": {
              "reasoningEffort": "xhigh",
              "reasoningSummary": "detailed",
              "textVerbosity": "medium"
            }
          }
        },
        "gpt-5.1-codex": {
          "name": "GPT 5.1 Codex (OAuth)",
          "limit": {
            "context": 272000,
            "output": 128000
          },
          "modalities": {
            "input": [
              "text",
              "image"
            ],
            "output": [
              "text"
            ]
          },
          "variants": {
            "low": {
              "reasoningEffort": "low",
              "reasoningSummary": "auto",
              "textVerbosity": "medium"
            },
            "medium": {
              "reasoningEffort": "medium",
              "reasoningSummary": "auto",
              "textVerbosity": "medium"
            },
            "high": {
              "reasoningEffort": "high",
              "reasoningSummary": "detailed",
              "textVerbosity": "medium"
            }
          }
        },
        "gpt-5.1-codex-mini": {
          "name": "GPT 5.1 Codex Mini (OAuth)",
          "limit": {
            "context": 272000,
            "output": 128000
          },
          "modalities": {
            "input": [
              "text",
              "image"
            ],
            "output": [
              "text"
            ]
          },
          "variants": {
            "medium": {
              "reasoningEffort": "medium",
              "reasoningSummary": "auto",
              "textVerbosity": "medium"
            },
            "high": {
              "reasoningEffort": "high",
              "reasoningSummary": "detailed",
              "textVerbosity": "medium"
            }
          }
        },
        "gpt-5.1": {
          "name": "GPT 5.1 (OAuth)",
          "limit": {
            "context": 272000,
            "output": 128000
          },
          "modalities": {
            "input": [
              "text",
              "image"
            ],
            "output": [
              "text"
            ]
          },
          "variants": {
            "none": {
              "reasoningEffort": "none",
              "reasoningSummary": "auto",
              "textVerbosity": "medium"
            },
            "low": {
              "reasoningEffort": "low",
              "reasoningSummary": "auto",
              "textVerbosity": "low"
            },
            "medium": {
              "reasoningEffort": "medium",
              "reasoningSummary": "auto",
              "textVerbosity": "medium"
            },
            "high": {
              "reasoningEffort": "high",
              "reasoningSummary": "detailed",
              "textVerbosity": "high"
            }
          }
        }
      }
    },
    "openrouter": {
      "models": {
        "kwaipilot/kat-coder-pro:free": {
          "name": "KAT-Coder-Pro (OpenRouter Free)"
        },
        "mistralai/devstral-2512:free": {
          "name": "Devstral 2 (OpenRouter Free)"
        },
        "deepseek/deepseek-r1:free": {
          "name": "DeepSeek R1 (OpenRouter Free)"
        },
        "meta-llama/llama-3.3-70b-instruct:free": {
          "name": "Llama 3.3 70B (OpenRouter Free)"
        },
        "qwen/qwen3-coder:free": {
          "name": "Qwen3 Coder (OpenRouter Free)"
        }
      }
    },
  },
  "$schema": "https://opencode.ai/config.json"
}

@cryptobetEU, see https://opencode.ai/docs/modes/#json-configuration

@cryptobetEU

Actually, looking at your config again - you already have opencode-anthropic-auth@0.0.8 installed.

Easiest fix for you: Just change @0.0.8 to @latest:

- "opencode-anthropic-auth@0.0.8",
+ "opencode-anthropic-auth@latest",

That's it. No need to add the mode block - the plugin does the same thing under the hood.

Hope that helps you :)

The best temporary fix is to add the following to ~/.config/opencode/opencode.json:

{
  "$schema": "https://opencode.ai/config.json",
  "plugin": [
    "opencode-anthropic-auth@latest"
  ]
}

@thamrx

@thamrx does this make it match claude code exact system prompt and tooling?

For me, it works even with the integrated oauth. I do not need to add anything into my opencode.json
however, I realized that opencode burns my claude max tokens in a way that I saw the first time the wait some hours message.
When I work with Claude Code, the tokens are much less and I basically have multible instances of it in parallel running on different projects and I never reach that limit.
Is anyone else experiencing opencode also as token burner?

@Cyb3rDudu

@Cyb3rDudu i also have noticed the same issue. It burns to my limit very quickly. I thought that was due to get-shit-done plugin, but seems i’m not the only one to experience a higher than normal usage of tokens

I believe Claude Code switches between Haiku/Opus automatically for efficiency. In OpenCode I can't see any way to use the automatic mode, if it's using Opus 4.5 for everything that would explain why it burns usage so quickly.

I believe Claude Code switches between Haiku/Opus automatically for efficiency. In OpenCode I can't see any way to use the automatic mode, if it's using Opus 4.5 for everything that would explain why it burns usage so quickly.

Claude code has their Explore agent specifically use Haiku, along with their statusline-setup using Sonnet and Claude Code Guide using Haiku, everything else inherits.

Opencode doesn't have set models, so you need to set up your subagents yourself(the builtin ones inherit)

@Cyb3rDudu

I believe Claude Code switches between Haiku/Opus automatically for efficiency. In OpenCode I can't see any way to use the automatic mode, if it's using Opus 4.5 for everything that would explain why it burns usage so quickly.

@Cyb3rDudu @ilbarone87 @cedws
You need to set in the opencode.json file:

  “small_model”: “anthropic/claude-haiku-4-5”,

  "agent": {
    "general": {
      "model": "anthropic/claude-haiku-4-5"
    },
    "explore": {
      "model": "anthropic/claude-haiku-4-5"
    },
  }

You can set any model, or even a different provider, as OpenRouter:

  "small_model": "openrouter/x-ai/grok-4.1-fast",

  "agent": {
    "general": {
      "model": "openrouter/x-ai/grok-4.1-fast"
    },
    "explore": {
      "model": "openrouter/x-ai/grok-4.1-fast"
    },
  }

I believe Claude Code switches between Haiku/Opus automatically for efficiency. In OpenCode I can't see any way to use the automatic mode, if it's using Opus 4.5 for everything that would explain why it burns usage so quickly.

@Cyb3rDudu @ilbarone87 @cedws You need to set in the opencode.json file:
  “small_model”: “anthropic/claude-haiku-4-5”,

  "agent": {
    "general": {
      "model": "anthropic/claude-haiku-4-5"
    },
    "explore": {
      "model": "anthropic/claude-haiku-4-5"
    },
  }
You can set any model, or even a different provider, as OpenRouter:
  "small_model": "openrouter/x-ai/grok-4.1-fast",

  "agent": {
    "general": {
      "model": "openrouter/x-ai/grok-4.1-fast"
    },
    "explore": {
      "model": "openrouter/x-ai/grok-4.1-fast"
    },
  }

I use google gemini antigravity high for explore and general

@Cyb3rDudu

@Cyb3rDudu @ilbarone87 its totally expected behaviour. native claude code is "dumber" partly because it relies more on cached tokens. and it is the reason why they from time to time ban max accounts using opencode. If Im not mistaken API (not pro/max sub) has control for caching.

Navigation Menu

Repository navigation

Broken Claude Max #7410

Description

Plugins

OpenCode version

Steps to reproduce

Screenshot and/or share link

Operating System

Terminal

360 remaining items

✅ Workaround confirmed working

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Broken Claude Max #7410

Description

Description

Plugins

OpenCode version

Steps to reproduce

Screenshot and/or share link

Operating System

Terminal

Activity

github-actions commented on Jan 9, 2026

iloveicedgreentea commented on Jan 9, 2026

Manokit commented on Jan 9, 2026

azizikri commented on Jan 9, 2026

Naomarik commented on Jan 9, 2026

fm1randa commented on Jan 9, 2026

diasm3 commented on Jan 9, 2026

andrhlt commented on Jan 9, 2026

arthur404dev commented on Jan 9, 2026

piotryordanov commented on Jan 9, 2026

kandotrun commented on Jan 9, 2026

thdxr commented on Jan 9, 2026

360 remaining items

skhaz commented on Jan 28, 2026

AceRothstein71 commented on Jan 28, 2026

✅ Workaround confirmed working

thamrx commented on Jan 28, 2026

Steffen025 commented on Jan 28, 2026

thamrx commented on Jan 28, 2026

johnmalek312 commented on Jan 29, 2026

Cyb3rDudu commented on Jan 30, 2026

ilbarone87 commented on Jan 30, 2026

cedws commented on Feb 1, 2026

Christopher-R-Perkins commented on Feb 1, 2026

cantalupo555 commented on Feb 2, 2026

AceRothstein71 commented on Feb 2, 2026

dearkafka commented on Feb 6, 2026

SpencerWell commented on May 1, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions