Skip to content

Set gpt-5.6-luna as default OpenAI model; update OpenAI and Claude pricing #319

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
jcsalterego wants to merge 3 commits into
base: main
Choose a base branch
from
Open

Set gpt-5.6-luna as default OpenAI model; update OpenAI and Claude pricing #319

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
b667563
Set gpt-5.6-luna as the default openai provider model
jcsalterego Jul 30, 2026
f9b5ed3
Add OpenAI gpt-5.6 pricing tier and fill in the missing gpt-5.4-nano …
jcsalterego Jul 30, 2026
0b5707b
Update Claude provider pricing to current published rates
jcsalterego Jul 30, 2026
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
4 changes: 2 additions & 2 deletions docs/settings-reference.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -148,7 +148,7 @@ provider's model, command and API-key variable separate.
"provider": "openai",
"providerConfigs": {
"openai": {
"model": "gpt-5.4-nano"
"model": "gpt-5.6-luna"
},
"claude-api": {
"model": "claude-haiku-4-5"
Expand Down Expand Up @@ -181,7 +181,7 @@ Provider defaults:
| `claude-cli` | `haiku` | None | `model`, `claudeCliPath`, `effort` |
| `command` | None | None | `command` |
| `claude-api` | `claude-haiku-4-5` | `ANTHROPIC_API_KEY` | `model`, `apiKeyEnv`, `maxTokens`, `effort` |
| `openai` | `gpt-5.4-nano` | `OPENAI_API_KEY` | `model`, `baseUrl`, `apiKeyEnv`, `maxTokens`, `effort` |
| `openai` | `gpt-5.6-luna` | `OPENAI_API_KEY` | `model`, `baseUrl`, `apiKeyEnv`, `maxTokens`, `effort` |
| `gemini` | `gemini-3.1-flash-lite` | `GEMINI_API_KEY` | `model`, `apiKeyEnv`, `maxTokens`, `effort` |
| `openrouter` | None | `OPENROUTER_API_KEY` | `model`, `apiKeyEnv`, `maxTokens`, `effort` |
| `off` | None | None | None |
Expand Down
2 changes: 1 addition & 1 deletion src/llm/providers/openai.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@ import { openaiCompatibleComplete } from "./openai-compatible.ts";
import type { ProviderCall, ProviderDescriptor } from "../types.ts";

export const DEFAULT_OPENAI_BASE_URL = "https://api.openai.com/v1";
export const DEFAULT_OPENAI_MODEL = "gpt-5.4-nano";
export const DEFAULT_OPENAI_MODEL = "gpt-5.6-luna";

export const openaiProvider: ProviderDescriptor = {
name: "openai",
Expand Down
38 changes: 34 additions & 4 deletions src/pricing.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,12 +16,24 @@ export interface Price {
// OpenAI/Codex and Gemini cached input is represented in the shared cacheRead bucket.
// Override any of these via $ARGUS_CONFIG_DIR/pricing.json.
const DEFAULTS: Record<string, Price> = {
opus: { input: 15, output: 75, cacheRead: 1.5, cacheWrite5m: 18.75, cacheWrite1h: 30 },
fable: { input: 10, output: 50, cacheRead: 1, cacheWrite5m: 12.5, cacheWrite1h: 20 },
mythos: { input: 10, output: 50, cacheRead: 1, cacheWrite5m: 12.5, cacheWrite1h: 20 },
opus: { input: 5, output: 25, cacheRead: 0.5, cacheWrite5m: 6.25, cacheWrite1h: 10 },
// Opus 4.1 (deprecated) and Opus 4 (retired) kept their pre-Opus-4.5 rate.
"opus-legacy": { input: 15, output: 75, cacheRead: 1.5, cacheWrite5m: 18.75, cacheWrite1h: 30 },
// Sonnet 5 introductory pricing runs through 2026-08-31, then reverts to the `sonnet` rate below.
"sonnet-5": { input: 2, output: 10, cacheRead: 0.2, cacheWrite5m: 2.5, cacheWrite1h: 4 },
sonnet: { input: 3, output: 15, cacheRead: 0.3, cacheWrite5m: 3.75, cacheWrite1h: 6 },
haiku: { input: 1, output: 5, cacheRead: 0.1, cacheWrite5m: 1.25, cacheWrite1h: 2 },
// Haiku 3.5, retired except on Bedrock/Google Cloud.
"haiku-legacy": { input: 0.8, output: 4, cacheRead: 0.08, cacheWrite5m: 1, cacheWrite1h: 1.6 },
"gpt-5.6": { input: 5, output: 30, cacheRead: 0.5, cacheWrite5m: 0, cacheWrite1h: 0 },
"gpt-5.6-terra": { input: 2, output: 12, cacheRead: 0.2, cacheWrite5m: 0, cacheWrite1h: 0 },
"gpt-5.6-luna": { input: 0.2, output: 1.2, cacheRead: 0.02, cacheWrite5m: 0, cacheWrite1h: 0 },
"gpt-5.5": { input: 5, output: 30, cacheRead: 0.5, cacheWrite5m: 0, cacheWrite1h: 0 },
"gpt-5.4": { input: 2.5, output: 15, cacheRead: 0.25, cacheWrite5m: 0, cacheWrite1h: 0 },
"gpt-5.4-mini": { input: 0.75, output: 4.5, cacheRead: 0.075, cacheWrite5m: 0, cacheWrite1h: 0 },
"gpt-5.4-nano": { input: 0.2, output: 1.25, cacheRead: 0.02, cacheWrite5m: 0, cacheWrite1h: 0 },
"gpt-5.3": { input: 1.75, output: 14, cacheRead: 0.175, cacheWrite5m: 0, cacheWrite1h: 0 },
"gpt-5": { input: 1.25, output: 10, cacheRead: 0.125, cacheWrite5m: 0, cacheWrite1h: 0 },
"codex-mini": { input: 1.5, output: 6, cacheRead: 0.375, cacheWrite5m: 0, cacheWrite1h: 0 },
Expand All @@ -48,11 +60,29 @@ const unpriced = new Set<string>();
/** Resolve a model id to its price family, or null if unknown / synthetic. */
function priceFor(model: string, usage?: Usage): Price | null {
const m = model.toLowerCase();
if (m.includes("opus")) return table.opus!;
if (m.includes("sonnet")) return table.sonnet!;
if (m.includes("haiku")) return table.haiku!;
if (m.includes("fable")) return table.fable!;
if (m.includes("mythos")) return table.mythos!;
if (m.includes("opus")) {
// Opus 4.1 (e.g. "claude-opus-4-1-...") and bare Opus 4 (e.g. "claude-opus-4-20250514")
// kept the older, higher rate; Opus 4.5 and later (incl. "opus-5") share the current one.
if (/opus-4-1(?!\d)/.test(m) || /opus-4(?:$|-\d{8})/.test(m)) return table["opus-legacy"]!;
return table.opus!;
}
if (m.includes("sonnet")) {
// Sonnet 5 (not 4.x) is on introductory pricing through 2026-08-31.
if (/sonnet-5(?:$|-)/.test(m)) return table["sonnet-5"]!;
return table.sonnet!;
}
if (m.includes("haiku")) {
if (m.includes("haiku-3") || m.includes("haiku-3.5")) return table["haiku-legacy"]!;
return table.haiku!;
}
if (m.includes("codex-mini")) return table["codex-mini"]!;
if (m.includes("gpt-5.6-luna")) return table["gpt-5.6-luna"]!;
if (m.includes("gpt-5.6-terra")) return table["gpt-5.6-terra"]!;
if (m.includes("gpt-5.6")) return table["gpt-5.6"]!;
if (m.includes("gpt-5.5")) return table["gpt-5.5"]!;
if (m.includes("gpt-5.4-nano")) return table["gpt-5.4-nano"]!;
if (m.includes("gpt-5.4-mini") || m.includes("gpt-5.4 mini")) return table["gpt-5.4-mini"]!;
if (m.includes("gpt-5.4")) return table["gpt-5.4"]!;
if (m.includes("gpt-5.3") || m.includes("gpt-5.2")) return table["gpt-5.3"]!;
Expand Down
4 changes: 2 additions & 2 deletions test/dashboard-views.test.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -58,8 +58,8 @@ describe("usage builders", () => {
];
const { bySource } = buildUsageBySource(rows, [{ source: "claude", sessions: 3 }]);
const claude = bySource.find((s) => s.name === "claude")!;
// opus input 15/Mtok + haiku input 1/Mtok, priced separately then summed.
expect(claude.cost).toBeCloseTo((1000 * 15 + 1000 * 1) / 1e6, 9);
// opus input 5/Mtok + haiku input 1/Mtok, priced separately then summed.
expect(claude.cost).toBeCloseTo((1000 * 5 + 1000 * 1) / 1e6, 9);
expect(claude.meta?.sessions).toBe(3);
});

Expand Down
15 changes: 14 additions & 1 deletion test/pricing.test.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,10 +6,19 @@ const z = { input: 0, output: 0, cacheRead: 0, cacheWrite5m: 0, cacheWrite1h: 0
describe("cost", () => {
test("prices 1M input tokens per model family at list rates", () => {
expect(cost({ ...z, input: 1_000_000 }, "claude-sonnet-4-6")).toBeCloseTo(3, 6);
expect(cost({ ...z, input: 1_000_000 }, "claude-opus-4-8")).toBeCloseTo(15, 6);
expect(cost({ ...z, input: 1_000_000 }, "claude-opus-4-8")).toBeCloseTo(5, 6);
expect(cost({ ...z, input: 1_000_000 }, "claude-haiku-4-5-20251001")).toBeCloseTo(1, 6);
});

test("prices legacy/deprecated Claude tiers and the newest models at their own rates", () => {
expect(cost({ ...z, input: 1_000_000 }, "claude-opus-4-1-20250805")).toBeCloseTo(15, 6);
expect(cost({ ...z, input: 1_000_000 }, "claude-opus-4-20250514")).toBeCloseTo(15, 6);
expect(cost({ ...z, input: 1_000_000 }, "claude-opus-5")).toBeCloseTo(5, 6);
expect(cost({ ...z, input: 1_000_000 }, "claude-sonnet-5")).toBeCloseTo(2, 6);
expect(cost({ ...z, input: 1_000_000 }, "claude-haiku-3-5-20241022")).toBeCloseTo(0.8, 6);
expect(cost({ ...z, input: 1_000_000 }, "claude-fable-5")).toBeCloseTo(10, 6);
});

test("prices output and cache classes", () => {
expect(cost({ ...z, output: 1_000_000 }, "claude-sonnet-4-6")).toBeCloseTo(15, 6);
expect(cost({ ...z, cacheRead: 1_000_000 }, "claude-sonnet-4-6")).toBeCloseTo(0.3, 6);
Expand All @@ -27,6 +36,10 @@ describe("cost", () => {
expect(cost({ ...z, output: 1_000_000 }, "gpt-5.4-mini")).toBeCloseTo(4.5, 6);
expect(cost({ ...z, input: 1_000_000 }, "gpt-5.3-codex")).toBeCloseTo(1.75, 6);
expect(cost({ ...z, cacheRead: 1_000_000 }, "codex-mini-latest")).toBeCloseTo(0.375, 6);
expect(cost({ ...z, input: 1_000_000 }, "gpt-5.6-luna")).toBeCloseTo(0.2, 6);
expect(cost({ ...z, output: 1_000_000 }, "gpt-5.6-terra")).toBeCloseTo(12, 6);
expect(cost({ ...z, input: 1_000_000 }, "gpt-5.6")).toBeCloseTo(5, 6);
expect(cost({ ...z, output: 1_000_000 }, "gpt-5.4-nano")).toBeCloseTo(1.25, 6);
});

test("prices Gemini models and the 2.5 Pro long-context tier", () => {
Expand Down