Repository navigation
Add Claude Mythos 5 and 5.1 pricing - #996
Conversation
Add claude-mythos-5 and claude-mythos-5-1 to the built-in Claude pricing table. Both bill input at 10 and output at 50 per million tokens with a 12.50 cache write; Mythos 5.1 cache reads are 0.25 per million, Mythos 5 cache reads are 1.00 per million. Mythos is absent from the models.dev catalog, so the built-in rows are what prices it. Upstream report-semantics version bump and predecessor parser hash are macOS-specific and not ported.
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info
📝 Walkthrough
Merge Risk: ⚪ Minimal · up to
|
|
@coderabbitai review |
|
|
@coderabbitai review |
✅ Action performedReview finished.
|
Port of the pricing part of upstream steipete/CodexBar steipete#4578 (
4654968e9) to Win-CodexBar.Branch:
port/claude-mythos-pricing(base96d5531e4), local commit7e626fcbb04f6b9e40a416a3fc7070b86cedccdb. Not pushed.Behavior
rust/src/core/cost_pricing.rs,CLAUDE_PRICING):claude-mythos-5: input 1e-5, output 5e-5, cache write 1.25e-5, cache read 1e-6 (USD per token; $10 / $50 / $12.50 / $1.00 per million).claude-mythos-5-1: same, but cache read 2.5e-7 ($0.25 per million).Not ported (macOS-specific):
Files
rust/src/core/cost_pricing.rsrust/src/core/cost_pricing_tests.rs:prices_claude_modelsextended;claude_pricing_rowssnapshot extended.rust/src/cost_scanner/tests/claude_rates.rs: new testclaude_mythos_models_price_from_the_builtin_table.No docs change: no Windows doc lists per-model prices.
Tests
claude_mythos_models_price_from_the_builtin_table: cache-read price 1.00 vs 0.25 per million, 20 tokens, both models.prices_claude_models: mythos-5 and mythos-5-1 resolve to 0.035 expected cost.claude_pricing_rowssnapshot: both rows pinned.Checks (local, Windows)
cargo fmt --all: clean.cargo clippy --manifest-path rust/Cargo.toml --all-targets -- -D warnings: clean.cargo test --manifest-path rust/Cargo.toml --lib cost_pricing: 41 passed.cargo test --manifest-path rust/Cargo.toml --lib mythos: 1 passed.cargo test --manifest-path rust/Cargo.toml --lib: 5280 passed, 0 failed, 2 ignored.cargo clippy --manifest-path apps/desktop-tauri/src-tauri/Cargo.toml --all-targets -- -D warnings: clean.cargo test --manifest-path apps/desktop-tauri/src-tauri/Cargo.toml: 804 passed, 0 failed.pnpm testnot run for this branch.Suggested fake-data proof
Write a local Claude transcript (synthetic) with
claude-mythos-5-1assistant usage rows: 1,000,000 input, 1,000,000 output, 1,000,000 cache-read tokens. Expected cost: $10 + $50 + $0.25 = $60.25. Seeing a non-zero cost in the cost history for that model confirms the built-in row is used.Summary by CodeRabbit