Skip to content

feat(usage): 更新 GPT-6 Astra 与 GPT-5.6 模型价格 #46

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
ApoclyReol wants to merge 1 commit into
base: main
Choose a base branch
from
Open

feat(usage): 更新 GPT-6 Astra 与 GPT-5.6 模型价格 #46

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
37 changes: 29 additions & 8 deletions Sources/CodexUsageWidget/main.swift
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3668,13 +3668,26 @@ func estimateStaticTokens(_ text: String) -> Int64 {
private func modelTokenPrice(for model: String?) -> ModelTokenPrice {
let normalized = (model ?? "").lowercased()

if normalized.contains("gpt-6-astra") {
return ModelTokenPrice(
model: "gpt-6-astra",
inputPerMillion: 10,
cachedInputPerMillion: 1,
outputPerMillion: 50,
cacheWriteInputPerMillion: 12.5,
fastModeMultiplier: 2.5,
longContextInputMultiplier: 2,
longContextOutputMultiplier: 1.5,
usesReferencePricing: false
)
}
if normalized.contains("gpt-5.6-sol") || normalized == "gpt-5.6" {
return ModelTokenPrice(
model: "gpt-5.6-sol",
inputPerMillion: 5,
cachedInputPerMillion: 0.5,
outputPerMillion: 30,
cacheWriteInputPerMillion: 6.25,
inputPerMillion: 4,
cachedInputPerMillion: 0.4,
outputPerMillion: 20,
cacheWriteInputPerMillion: 5,
fastModeMultiplier: 2,
longContextInputMultiplier: 2,
longContextOutputMultiplier: 1.5,
Expand Down Expand Up @@ -3882,6 +3895,7 @@ private enum ModelPricingSelfTest {
totalTokens: 110_000
)
let sol = modelTokenPrice(for: "gpt-5.6")
let astra = modelTokenPrice(for: "gpt-6-astra-2026-04-30")
let terra = modelTokenPrice(for: "gpt-5.6-terra-2026-02-16")
let luna = modelTokenPrice(for: "GPT-5.6-LUNA")
let gpt55 = modelTokenPrice(for: "gpt-5.5")
Expand All @@ -3892,9 +3906,12 @@ private enum ModelPricingSelfTest {

expect(sol.model == "gpt-5.6-sol", "gpt-5.6 should resolve to gpt-5.6-sol")
expect(!sol.usesReferencePricing, "gpt-5.6 should use an explicit price")
expect(astra.model == "gpt-6-astra", "Astra snapshots should preserve the Astra price")
expect(!astra.usesReferencePricing, "gpt-6-astra should use an explicit price")
expect(terra.model == "gpt-5.6-terra", "terra snapshots should preserve the terra price")
expect(luna.model == "gpt-5.6-luna", "luna matching should be case-insensitive")
expect(nearlyEqual(estimatedCostUSD(tokens: sampleTokens, price: sol), 0.62), "Sol cached input estimate should use the split rates")
expect(nearlyEqual(estimatedCostUSD(tokens: sampleTokens, price: astra), 1.14), "Astra should use the official standard API rates")
expect(nearlyEqual(estimatedCostUSD(tokens: sampleTokens, price: sol), 0.456), "Sol should use the official standard API rates")
expect(nearlyEqual(estimatedCostUSD(tokens: sampleTokens, price: terra), 0.248), "Terra should use the official standard API rates")
expect(nearlyEqual(estimatedCostUSD(tokens: sampleTokens, price: luna), 0.0248), "Luna should use the official standard API rates")
expect(nearlyEqual(gpt55.inputPerMillion, 5) && nearlyEqual(gpt55.cachedInputPerMillion, 0.5) && nearlyEqual(gpt55.outputPerMillion, 30), "GPT-5.5 should use the official standard API rates")
Expand All @@ -3911,8 +3928,10 @@ private enum ModelPricingSelfTest {
reasoningOutputTokens: 0,
totalTokens: 110_000
)
expect(nearlyEqual(estimatedCostUSD(tokens: cacheWriteTokens, price: sol), 0.6325), "GPT-5.6 cache writes should use the 1.25x write rate")
expect(nearlyEqual(estimatedCostUSD(tokens: cacheWriteTokens, price: sol, serviceTier: "priority"), 1.265), "GPT-5.6 Fast mode should use the published 2x API rates")
expect(nearlyEqual(estimatedCostUSD(tokens: cacheWriteTokens, price: astra), 1.165), "Astra cache writes should use the published write rate")
expect(nearlyEqual(estimatedCostUSD(tokens: cacheWriteTokens, price: sol), 0.466), "GPT-5.6 cache writes should use the published write rate")
expect(nearlyEqual(estimatedCostUSD(tokens: cacheWriteTokens, price: astra, serviceTier: "priority"), 2.9125), "Astra Fast mode should use the published 2.5x API rates")
expect(nearlyEqual(estimatedCostUSD(tokens: cacheWriteTokens, price: sol, serviceTier: "priority"), 0.932), "GPT-5.6 Fast mode should use the published 2x API rates")
expect(nearlyEqual(estimatedCostUSD(tokens: cacheWriteTokens, price: gpt55, serviceTier: "fast"), 1.55), "GPT-5.5 Fast mode should use the published 2.5x API rates")

let longContextTokens = TokenBreakdown(
Expand All @@ -3923,8 +3942,10 @@ private enum ModelPricingSelfTest {
reasoningOutputTokens: 0,
totalTokens: 400_000
)
expect(nearlyEqual(estimatedCostUSD(tokens: longContextTokens, price: sol), 6.85), "GPT-5.6 long context should use 2x input and 1.5x output rates")
expect(nearlyEqual(estimatedCostUSD(tokens: longContextTokens, price: astra), 12.2), "Astra long context should use 2x input and 1.5x output rates")
expect(nearlyEqual(estimatedCostUSD(tokens: longContextTokens, price: sol), 4.88), "GPT-5.6 long context should use 2x input and 1.5x output rates")
expect(isFastServiceTier("priority") && isFastServiceTier("FAST") && !isFastServiceTier("default"), "service tier normalization should distinguish Fast mode")
expect(!modelUsageUsesReferencePricing("gpt-6-astra"), "known Astra models should not use reference pricing")
expect(!modelUsageUsesReferencePricing("gpt-5.6-luna"), "known GPT-5.6 models should not use reference pricing")
expect(modelUsageUsesReferencePricing("gpt-5.3-codex-spark"), "Codex Spark should not inherit GPT-5.3 Codex pricing")
expect(modelUsageUsesReferencePricing("future-model"), "unknown models should retain reference pricing")
Expand Down
Loading