Error: state snapshot was created by a newer version

TL;DR: The state was last written by a newer tofu (or terraform) than the one you're running. Upgrade your binary to at least that version. Never hand-edit the version marker in the state file to force it to load.

Error: state snapshot was created by OpenTofu v1.12.0, which is newer than current v1.10.0

(The same error appears with Terraform version stamps when migrating between the two tools.)

Steps

1. Read both versions from the error: the writer (v1.12.0) and your current (v1.10.0).

Expected: you know the minimum version you need.

1. Install a binary at least as new as the writer: tofuenv install 1.12.6 && tofuenv use 1.12.6 (or tenv, or your OS package).

Expected: tofu version shows >= the writer version.

1. Re-run your command.

Expected: the state loads; the error is gone.

When this applies

• The error names two versions and yours is older.
• A teammate or CI applied with a newer release than you have installed.
• During Terraform-to-OpenTofu migration, when the state was last written by a Terraform newer than your tofu.

When it doesn't apply

• Error: Unsupported OpenTofu Core version is about your CONFIG's required_version, not the state's writer version.
• Corrupt or truncated state files fail with parse errors, not version errors.

Tool versions

All OpenTofu versions. State format v4 is shared across recent Terraform and OpenTofu, so the check keys on the version stamp, not the tool name.

Why it happens

State records the exact version that last wrote it. A newer writer may have used state features the older reader doesn't understand, so the older tool refuses rather than risk misreading your infrastructure.

Edge cases

• Never hand-edit version or terraform_version in the state to force a load. You risk corrupting resource decoding and losing the state you were trying to save.
• Teams should pin one tofu version (.opentofu-version file, CI matrix, required_version) so nobody leapfrogs the rest.
• During migration, treat the cutover as one-way: add a CI guard so nobody keeps running terraform apply against tofu-managed state and stamps the marker back.