From 773c9e374371e9b43a4bba177d86afb3b57b8aaf Mon Sep 17 00:00:00 2001 From: pierreraby Date: Tue, 17 Nov 2026 10:09:51 +0100 Subject: [PATCH] chore(models): refresh November 17 catalog and Flint pricing Sync command-code 1.89.2 capabilities and review the 101-model API snapshot. Add undated free-tier pricing for Flint and explicit model/routing and pricing-page ID regressions. Correct the Unreleased description following #220. --- CHANGELOG.md | 3 +- README.md | 2 +- src/commandcode-catalog.ts | 11 ++++- src/pricing.ts | 4 +- tests/fixtures/commandcode-model-ids.json | 3 +- tests/fixtures/commandcode-pricing.json | 3 +- tests/test-models.ts | 57 +++++++++++++++++++++++ tests/test-pricing-check.ts | 9 ++++ tests/test-pricing.ts | 21 ++++++++- 9 files changed, 104 insertions(+), 9 deletions(-) diff --git a/CHANGELOG.md b/CHANGELOG.md index c2ffd5f..0c6e06a 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -2,7 +2,8 @@ ## Unreleased -- Add reviewed display pricing for the two Provider API additions published after the previous snapshot: `Qwen/Qwen4-Max` at $1.50 input / $4.50 output / $0.075 cache read / $1.875 cache write per million tokens, and `tencent/hy4-flash` at $0.07 / $0.281 / $0.003 with no cache-write charge. Both values come from the official pricing page rows `qwen-4-max` and `tencent/hy4-flash`, publish no long-context tier, and carry no promotion. No model left the API in this window. Refresh the 100-model API and pricing snapshots to 2026-11-16; the pricing check reports no drift. Static capability metadata stays at `command-code@1.89.1`, which does not list either model, so both remain text-only without reasoning or effort flags until upstream ships them. +- Add reviewed display pricing for the two Provider API additions: `Qwen/Qwen4-Max` at $1.50 input / $4.50 output / $0.075 cache read / $1.875 cache write per million tokens, and `tencent/hy4-flash` at $0.07 / $0.281 / $0.003 with no cache-write charge. Both values come from the official pricing page rows `qwen-4-max` and `tencent/hy4-flash`, publish no long-context tier, and carry no promotion. ([#220](https://github.com/patlux/pi-commandcode-provider/pull/220)) +- Refresh the capability snapshot to `command-code@1.89.2`: advertise image input, reasoning and selectable efforts for `Qwen/Qwen4-Max` (`off`, `low`, `high`, `xhigh`) and `tencent/hy4-flash` (`off`, `high`). Add text-only reasoning for `recursive/flint-v1.0:free`, with no selectable effort metadata. All three retain the default 65,536-token output limit, capped by context size, and use Responses when advertised by the API. Add Flint's reviewed zero display prices while its undated free tier lasts, and refresh the 101-model API and pricing snapshots to 2026-11-17. No model left the API; existing prices, endpoint routing rules and model-specific output limits are unchanged. ## 0.8.4 - 2026-11-15 diff --git a/README.md b/README.md index 58c46b9..10dd46f 100644 --- a/README.md +++ b/README.md @@ -180,7 +180,7 @@ The following environment variables are intended for tests, local mocks, and com ## Image input -The provider advertises image input only for models marked with the `image` input modality in the official Command Code CLI model catalog. The capability snapshot currently follows `command-code@1.89.1`; unknown models default to text-only until their upstream metadata is reviewed. A daily GitHub Actions job synchronizes the CLI version, image capabilities, reasoning flags, reasoning efforts, and model-specific output limits with the latest published CLI package, also dropping manual effort overrides that upstream has published itself, and opens or updates a reviewable pull request when they change. Pricing remains manually reviewed because temporary promotions and long-context tiers require explicit review. +The provider advertises image input only for models marked with the `image` input modality in the official Command Code CLI model catalog. The capability snapshot currently follows `command-code@1.89.2`; unknown models default to text-only until their upstream metadata is reviewed. A daily GitHub Actions job synchronizes the CLI version, image capabilities, reasoning flags, reasoning efforts, and model-specific output limits with the latest published CLI package, also dropping manual effort overrides that upstream has published itself, and opens or updates a reviewable pull request when they change. Pricing remains manually reviewed because temporary promotions and long-context tiers require explicit review. For vision-capable models, Pi's native provider adapters forward image blocks from user messages and tool results using the documented OpenAI or Anthropic message schema. Unknown and text-only models remain marked text-only in Pi. diff --git a/src/commandcode-catalog.ts b/src/commandcode-catalog.ts index 7ddb481..5209f2c 100644 --- a/src/commandcode-catalog.ts +++ b/src/commandcode-catalog.ts @@ -1,4 +1,4 @@ -export const COMMAND_CODE_CLI_VERSION = "1.89.1" +export const COMMAND_CODE_CLI_VERSION = "1.89.2" export type CommandCodeInputType = "text" | "image" export type CommandCodeReasoningEffort = @@ -11,7 +11,7 @@ export type CommandCodeReasoningEffort = | "max" /** - * Generated from command-code@1.89.1 by `npm run sync:commandcode-catalog`. + * Generated from command-code@1.89.2 by `npm run sync:commandcode-catalog`. * Do not edit manually. */ export const MODEL_INPUT_MODALITIES: Readonly> = { @@ -79,9 +79,11 @@ export const MODEL_INPUT_MODALITIES: Readonly> = { "Qwen/Qwen3.8-Max-0902": true, "Qwen/Qwen3.8-Max-Prime": true, "Qwen/Qwen3.8-Omni-Flash": true, + "Qwen/Qwen4-Max": true, + "recursive/flint-v1.0:free": true, "sakana/fugu-ultra": true, "stealth/glyph-cluster:free": true, "stepfun/Step-3.5-Flash": true, "stepfun/Step-3.7-Flash": true, "stepfun/Step-5-Preview": true, "tencent/hy3-paid": true, + "tencent/hy4-flash": true, "tencent/hy4-preview": true, "thinkingmachines/inkling": true, "thinkingmachines/inkling-2-small:free": true, @@ -246,9 +251,11 @@ export const MODEL_EFFORTS: Readonly> = { // Free models + // Free while the free tier lasts; no fixed expiry is published. + "recursive/flint-v1.0:free": { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 }, // Free while the stealth preview lasts; no fixed expiry is published. "stealth/glyph-cluster:free": { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 }, // Free while the free tier lasts; no fixed expiry is published. diff --git a/tests/fixtures/commandcode-model-ids.json b/tests/fixtures/commandcode-model-ids.json index 9c82c01..4bfd20b 100644 --- a/tests/fixtures/commandcode-model-ids.json +++ b/tests/fixtures/commandcode-model-ids.json @@ -1,5 +1,5 @@ { - "fetchedAt": "2026-11-16T09:22:58.757Z", + "fetchedAt": "2026-11-17T08:30:20.471Z", "source": "https://api.commandcode.ai/provider/v1/models", "modelIds": [ "claude-sonnet-5-5", @@ -87,6 +87,7 @@ "thinkingmachines/inkling", "thinkingmachines/inkling-small", "thinkingmachines/inkling-2-small:free", + "recursive/flint-v1.0:free", "stealth/glyph-cluster:free", "inclusionai/ling-3.1-flash", "inclusionai/ling-3.1-flash-1019", diff --git a/tests/fixtures/commandcode-pricing.json b/tests/fixtures/commandcode-pricing.json index b8a7369..4c5a21b 100644 --- a/tests/fixtures/commandcode-pricing.json +++ b/tests/fixtures/commandcode-pricing.json @@ -1,5 +1,5 @@ { - "verifiedAt": "2026-11-16", + "verifiedAt": "2026-11-17", "source": "https://commandcode.ai/docs/resources/pricing-limits", "tierPolicy": "Use request-wide input tiers; the highest threshold exceeded by input plus cache tokens applies to the full request.", "tiers": { @@ -22,6 +22,7 @@ "xai/grok-4.7": [[200000, 4, 12, 1, 0]] }, "costs": { + "recursive/flint-v1.0:free": [0, 0, 0, 0], "stealth/glyph-cluster:free": [0, 0, 0, 0], "thinkingmachines/inkling-2-small:free": [0, 0, 0, 0], "inclusionai/ling-3.1-flash-1019": [0.3, 0.9, 0.06, 0], diff --git a/tests/test-models.ts b/tests/test-models.ts index 60e943a..d5a41a5 100644 --- a/tests/test-models.ts +++ b/tests/test-models.ts @@ -197,6 +197,63 @@ describe("commandCodeModelsFromApiResponse()", () => { } }) + for (const { id, name, contextLength, input, efforts } of [ + { + id: "Qwen/Qwen4-Max", + name: "Qwen 4 Max", + contextLength: 1_000_000, + input: ["text", "image"], + efforts: ["off", "low", "high", "xhigh"], + }, + { + id: "tencent/hy4-flash", + name: "Tencent Hy4 Flash", + contextLength: 1_024_000, + input: ["text", "image"], + efforts: ["off", "high"], + }, + { + id: "recursive/flint-v1.0:free", + name: "Flint", + contextLength: 262_144, + input: ["text"], + efforts: undefined, + }, + ]) { + it(`advertises the command-code@1.89.2 capabilities and API routing for ${id}`, () => { + assert.deepEqual(inputModalitiesForModel(id), input) + assert.equal(MODEL_REASONING[id], true) + assert.deepEqual(MODEL_EFFORTS[id], efforts) + assert.deepEqual(thinkingMetadataForModel(id)?.thinking?.efforts, efforts) + assert.ok(thinkingMetadataForModel(id), "reasoning metadata must be registered") + assert.equal(MODEL_MAX_OUTPUT_TOKENS[id], undefined, "use the default output limit") + for (const contextWindow of [contextLength, 16_384]) { + const models = commandCodeModelsFromApiResponse({ + object: "list", + data: [ + { + ...API_RESPONSE.data[0], + id, + name, + context_length: contextWindow, + supported_endpoints: ["/chat/completions", "/responses"], + }, + ], + }) + assert.deepEqual(models, [ + { + id, + name: `${name} (CC)`, + api: "openai-responses", + reasoning: true, + contextWindow, + maxTokens: Math.min(65_536, contextWindow), + }, + ]) + } + }) + } + it("advertises Qwen 3.8 Omni Flash capabilities from the CLI catalog", () => { const id = "Qwen/Qwen3.8-Omni-Flash" assert.deepEqual(inputModalitiesForModel(id), ["text", "image"]) diff --git a/tests/test-pricing-check.ts b/tests/test-pricing-check.ts index 930903f..790fc01 100644 --- a/tests/test-pricing-check.ts +++ b/tests/test-pricing-check.ts @@ -319,6 +319,15 @@ describe("pricing rate and context validation", () => { describe("pricing page id matching", () => { it("derives page id candidates from API ids and aliases", () => { + assert.deepEqual(pricingPageIdCandidates("Qwen/Qwen4-Max"), ["qwen/qwen4-max", "qwen-4-max"]) + assert.deepEqual(pricingPageIdCandidates("tencent/hy4-flash"), [ + "tencent/hy4-flash", + "hy4-flash", + ]) + assert.deepEqual(pricingPageIdCandidates("recursive/flint-v1.0:free"), [ + "recursive/flint-v1.0:free", + "flint-v1.0:free", + ]) assert.deepEqual(pricingPageIdCandidates("tencent/hy4-preview"), [ "tencent/hy4-preview", "hy4-preview", diff --git a/tests/test-pricing.ts b/tests/test-pricing.ts index 0e22e23..6c0f54b 100644 --- a/tests/test-pricing.ts +++ b/tests/test-pricing.ts @@ -27,7 +27,11 @@ const fixtureUrl = new URL("./fixtures/commandcode-model-ids.json", import.meta. const fixture = JSON.parse(await readFile(fixtureUrl, "utf-8")) as ModelCatalogSnapshot const pricingFixtureUrl = new URL("./fixtures/commandcode-pricing.json", import.meta.url) const pricingFixture = JSON.parse(await readFile(pricingFixtureUrl, "utf-8")) as PricingSnapshot -const freeModels = new Set(["stealth/glyph-cluster:free", "thinkingmachines/inkling-2-small:free"]) +const freeModels = new Set([ + "recursive/flint-v1.0:free", + "stealth/glyph-cluster:free", + "thinkingmachines/inkling-2-small:free", +]) function assertCost( modelId: string, @@ -282,6 +286,19 @@ describe("MODEL_COSTS pricing overlay", () => { }) }) + it("prices Flint as free during its undated free tier", () => { + assertCost("recursive/flint-v1.0:free", { + input: 0, + output: 0, + cacheRead: 0, + cacheWrite: 0, + }) + assert.ok( + !TEMPORARY_PRICING.some((pricing) => pricing.models.includes("recursive/flint-v1.0:free")), + "the source publishes no fixed expiry for this free tier", + ) + }) + it("prices Inkling 2 Small free as free during its undated free tier", () => { assertCost("thinkingmachines/inkling-2-small:free", { input: 0, @@ -475,7 +492,7 @@ describe("MODEL_COSTS pricing overlay", () => { it("tracks pricing provenance", () => { assert.equal(PRICING_SOURCE_URL, "https://commandcode.ai/docs/resources/pricing-limits") - assert.equal(PRICING_LAST_VERIFIED, "2026-11-16") + assert.equal(PRICING_LAST_VERIFIED, "2026-11-17") }) it("fails once temporary pricing needs review", () => {