prefix_binding_mismatch
Anthropic (Claude) input_transformations entry: type: "thinking_dropped", reason: "prefix_binding_mismatch".
What it means
Something before this thinking block (system, tools or an earlier message) changed after Claude produced it, so the API dropped the block. That happens with prefix_mismatch_behavior: "drop_block".
thinking_dropped,prefix_binding_mismatch: What came before this block isn’t what it recorded. Yours to fix
Fix a thinking block bound to a different conversation (checked 2026-11-20)
Is it yours to fix?
Yes. Your code changed something, so it’s yours to fix.
How to fix it
Send every earlier message exactly as you sent it before, and put each update in a new message at the end. Store any context your code adds to a message with that message.
An example that produces it (drop_block set (stopgap)):
{
"max_tokens": 4096,
"messages": [
{
"content": "Fix the failing test.",
"role": "user"
}
],
"model": "claude-fable-5-2",
"system": "You are a coding agent. Today is 2026-11-18."
}{
"betas": [
"thinking-binding-controls-2026-08-01"
],
"max_tokens": 4096,
"messages": [
{
"content": "Fix the failing test.",
"role": "user"
},
{
"content": [
{
"signature": "sigA1",
"thinking": "",
"type": "thinking"
},
{
"text": "Done.",
"type": "text"
}
],
"role": "assistant"
},
{
"content": "Now run the linter.",
"role": "user"
}
],
"model": "claude-fable-5-2",
"system": "You are a coding agent. Today is 2026-11-19.",
"thinking": {
"block_binding": {
"prefix_mismatch_behavior": "drop_block"
},
"type": "adaptive"
}
}Related reason codes
thinking_mismatch_allowed: thinking_mismatch_allowed: Claude read stale thinking and you were billed for itmodel_binding_mismatch: model_binding_mismatch: this Claude model can’t read thinking from the model that produced itorganization_binding_mismatch: organization_binding_mismatch: thinking from another account was dropped
Last checked 2026-11-20 against the official Anthropic docs.
FAQ
Does prefix_binding_mismatch cost me anything?
You aren’t billed for the dropped block, but Claude can spend more output tokens working out reasoning it can no longer read, and dropping it invalidates the prompt cache from the first dropped block onward. When something changes on every turn, the increase repeats.
How do I find what changed?
Compare system, tools and messages up to the first entry’s path with the conversation as it was when that block was produced. If every block is listed, start with system and tools. Or paste both requests into the prefix checker above.
Should my tests fail on this entry?
Yes. Anthropic’s test steps say to fail only on an entry whose reason is prefix_binding_mismatch; model and organization entries are expected after a switch.