@trazum/core 2.7.11 → 3.0.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +1 -1
- package/dist/advisories.d.ts +14 -7
- package/dist/advisories.d.ts.map +1 -1
- package/dist/advisories.js +27 -20
- package/dist/advisories.js.map +1 -1
- package/dist/budget.d.ts.map +1 -1
- package/dist/budget.js +2 -1
- package/dist/budget.js.map +1 -1
- package/dist/connector.js +1 -1
- package/dist/connector.js.map +1 -1
- package/dist/conversation.d.ts.map +1 -1
- package/dist/conversation.js +4 -2
- package/dist/conversation.js.map +1 -1
- package/dist/i18n/en.js +2 -2
- package/dist/i18n/en.js.map +1 -1
- package/dist/i18n/es.js +2 -2
- package/dist/i18n/es.js.map +1 -1
- package/dist/i18n/types.d.ts +2 -1
- package/dist/i18n/types.d.ts.map +1 -1
- package/dist/index.d.ts +3 -1
- package/dist/index.d.ts.map +1 -1
- package/dist/index.js +2 -1
- package/dist/index.js.map +1 -1
- package/dist/input-shape.d.ts.map +1 -1
- package/dist/input-shape.js +4 -2
- package/dist/input-shape.js.map +1 -1
- package/dist/openrouter.d.ts +1 -1
- package/dist/openrouter.d.ts.map +1 -1
- package/dist/openrouter.js +1 -2
- package/dist/openrouter.js.map +1 -1
- package/dist/phrases.d.ts +39 -0
- package/dist/phrases.d.ts.map +1 -1
- package/dist/phrases.js +130 -0
- package/dist/phrases.js.map +1 -1
- package/dist/pricing-overlay.d.ts +1 -1
- package/dist/pricing-overlay.d.ts.map +1 -1
- package/dist/pricing-overlay.js +21 -11
- package/dist/pricing-overlay.js.map +1 -1
- package/dist/pricing.d.ts +33 -6
- package/dist/pricing.d.ts.map +1 -1
- package/dist/pricing.js +77 -115
- package/dist/pricing.js.map +1 -1
- package/dist/repeats.d.ts.map +1 -1
- package/dist/repeats.js +4 -2
- package/dist/repeats.js.map +1 -1
- package/dist/reprice.js +1 -1
- package/dist/reprice.js.map +1 -1
- package/dist/review.js +2 -1
- package/dist/review.js.map +1 -1
- package/dist/rules.d.ts.map +1 -1
- package/dist/rules.js +119 -3
- package/dist/rules.js.map +1 -1
- package/dist/segment.d.ts +31 -1
- package/dist/segment.d.ts.map +1 -1
- package/dist/segment.js +128 -2
- package/dist/segment.js.map +1 -1
- package/dist/session-cost.d.ts.map +1 -1
- package/dist/session-cost.js +4 -2
- package/dist/session-cost.js.map +1 -1
- package/dist/session-ledger.d.ts.map +1 -1
- package/dist/session-ledger.js +4 -2
- package/dist/session-ledger.js.map +1 -1
- package/dist/structure.d.ts +34 -0
- package/dist/structure.d.ts.map +1 -1
- package/dist/structure.js +143 -12
- package/dist/structure.js.map +1 -1
- package/dist/truncation-retry.d.ts.map +1 -1
- package/dist/truncation-retry.js +4 -2
- package/dist/truncation-retry.js.map +1 -1
- package/dist/ttl-fit.d.ts.map +1 -1
- package/dist/ttl-fit.js +4 -2
- package/dist/ttl-fit.js.map +1 -1
- package/dist/types.d.ts +17 -7
- package/dist/types.d.ts.map +1 -1
- package/dist/usage.d.ts.map +1 -1
- package/dist/usage.js +13 -12
- package/dist/usage.js.map +1 -1
- package/dist/watch.d.ts +11 -1
- package/dist/watch.d.ts.map +1 -1
- package/dist/watch.js.map +1 -1
- package/dist/weekly.d.ts +140 -0
- package/dist/weekly.d.ts.map +1 -0
- package/dist/weekly.js +176 -0
- package/dist/weekly.js.map +1 -0
- package/dist/write.js +2 -2
- package/dist/write.js.map +1 -1
- package/package.json +1 -1
- package/src/advisories.ts +41 -25
- package/src/budget.ts +2 -1
- package/src/connector.ts +1 -1
- package/src/conversation.ts +4 -2
- package/src/i18n/en.ts +2 -2
- package/src/i18n/es.ts +2 -2
- package/src/i18n/types.ts +2 -1
- package/src/index.ts +4 -2
- package/src/input-shape.ts +4 -2
- package/src/openrouter.ts +1 -2
- package/src/phrases.ts +134 -0
- package/src/pricing-overlay.ts +23 -14
- package/src/pricing.ts +82 -120
- package/src/repeats.ts +4 -2
- package/src/reprice.ts +1 -1
- package/src/review.ts +1 -1
- package/src/rules.ts +135 -2
- package/src/segment.ts +131 -3
- package/src/session-cost.ts +4 -2
- package/src/session-ledger.ts +4 -2
- package/src/structure.ts +152 -13
- package/src/truncation-retry.ts +4 -2
- package/src/ttl-fit.ts +4 -2
- package/src/types.ts +18 -7
- package/src/usage.ts +13 -12
- package/src/watch.ts +12 -1
- package/src/weekly.ts +281 -0
- package/src/write.ts +2 -2
package/src/pricing.ts
CHANGED
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
import type { InferenceGeo, ModelPricing, PricingTier, ServiceTier, TierCondition } from './types.js';
|
|
1
|
+
import type { Capability, InferenceGeo, ModelPricing, PricingTier, ServiceTier, TierCondition } from './types.js';
|
|
2
2
|
|
|
3
3
|
/**
|
|
4
4
|
* Model and pricing catalogue (USD per million tokens).
|
|
@@ -122,10 +122,17 @@ export const PROVIDER_REVIEWED: Readonly<Record<string, string>> = Object.freeze
|
|
|
122
122
|
* target's rates. 2.5 Pro carries that as a second period; 2.5 Flash's text
|
|
123
123
|
* rates are 3.5 Flash-Lite's, so its price does not move. The Gemini 4
|
|
124
124
|
* family and 3.5 Flash-Lite are priced in their own rows, from the same
|
|
125
|
-
* read; the page's other models are still unpriced.
|
|
126
|
-
*
|
|
127
|
-
*
|
|
128
|
-
*
|
|
125
|
+
* read; the page's other models are still unpriced.
|
|
126
|
+
*
|
|
127
|
+
* Batch tabs read again 2026-11-17, for the cache price alone. Six of the
|
|
128
|
+
* nine rows print a batch cache price that is half the standard one, as
|
|
129
|
+
* the batch input is: 4 Carbon, Argon and Hydrogen, 3.8 and 3.6 Flash and
|
|
130
|
+
* 3.5 Flash-Lite. Three do not. 3.1 Pro Preview's reads "Same as Standard",
|
|
131
|
+
* $0.20 and $0.40, and 2.5 Pro's and 2.5 Flash's print the standard figures
|
|
132
|
+
* ($0.125 and $0.25; $0.03). Those three carry `batchCacheRead: 0.2`: the
|
|
133
|
+
* cache price over the batch input, where a standard call reads at 0.1.
|
|
134
|
+
* For the two 2.5 models it ends on 2026-12-01, when they are billed at
|
|
135
|
+
* the rates of 3.8 Flash and 3.5 Flash-Lite, whose Batch tabs halve it.
|
|
129
136
|
*/
|
|
130
137
|
google: '2026-11-16',
|
|
131
138
|
/*
|
|
@@ -236,25 +243,39 @@ export const COST_MULTIPLIERS = {
|
|
|
236
243
|
* should fall back to the default.
|
|
237
244
|
*/
|
|
238
245
|
/** Memoised for the same reason as `effectivePricing`: once per record. */
|
|
239
|
-
|
|
240
|
-
|
|
241
|
-
|
|
242
|
-
>();
|
|
246
|
+
type Multipliers = { cacheWrite5m: number; cacheWrite1h: number; cacheRead: number; batch: number | null };
|
|
247
|
+
const multipliersMemo = new WeakMap<ModelPricing, Multipliers>();
|
|
248
|
+
const batchMultipliersMemo = new WeakMap<ModelPricing, Multipliers>();
|
|
243
249
|
|
|
244
|
-
|
|
245
|
-
|
|
246
|
-
|
|
247
|
-
|
|
248
|
-
|
|
249
|
-
|
|
250
|
+
/**
|
|
251
|
+
* The cache and batch ratios of a row, at a date and for a kind of call.
|
|
252
|
+
*
|
|
253
|
+
* `how` is the same context `effectivePricing` is given, and it matters for
|
|
254
|
+
* one figure. Every ratio here multiplies the input rate `effectivePricing`
|
|
255
|
+
* returned, which for a batch call is already the batch rate. Where a
|
|
256
|
+
* provider's Batch table charges the cache at something other than the
|
|
257
|
+
* standard share of that halved input, the row says so in `batchCacheRead`,
|
|
258
|
+
* and a batch call gets that as its `cacheRead`. Until 2.7.12 nothing asked
|
|
259
|
+
* how the call ran: Google's Batch tab for Gemini 3.1 Pro Preview and both
|
|
260
|
+
* 2.5 models prints "Same as Standard" for the cache, and a batch call's
|
|
261
|
+
* cache reads on them were billed at half of what the page charges.
|
|
262
|
+
*/
|
|
263
|
+
export function multipliersFor(model: ModelPricing, on?: Date, how?: { batch?: boolean }): Multipliers {
|
|
250
264
|
// A dated question about a repriced model asks the period in force then.
|
|
251
265
|
if (on !== undefined && model.history !== undefined) {
|
|
252
266
|
const period = pricedAt(model, on);
|
|
253
|
-
if (period !== model) return multipliersFor(period);
|
|
267
|
+
if (period !== model) return multipliersFor(period, undefined, how);
|
|
268
|
+
}
|
|
269
|
+
const m = model.multipliers ?? {};
|
|
270
|
+
if (how?.batch === true && m.batchCacheRead !== undefined && m.batch !== null) {
|
|
271
|
+
const cached = batchMultipliersMemo.get(model);
|
|
272
|
+
if (cached !== undefined) return cached;
|
|
273
|
+
const result = Object.freeze({ ...multipliersFor(model), cacheRead: m.batchCacheRead });
|
|
274
|
+
batchMultipliersMemo.set(model, result);
|
|
275
|
+
return result;
|
|
254
276
|
}
|
|
255
277
|
const cached = multipliersMemo.get(model);
|
|
256
278
|
if (cached !== undefined) return cached;
|
|
257
|
-
const m = model.multipliers ?? {};
|
|
258
279
|
const result = Object.freeze({
|
|
259
280
|
cacheWrite5m: m.cacheWrite5m ?? COST_MULTIPLIERS.cacheWrite5m,
|
|
260
281
|
cacheWrite1h: m.cacheWrite1h ?? COST_MULTIPLIERS.cacheWrite1h,
|
|
@@ -418,7 +439,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
418
439
|
*/
|
|
419
440
|
tiers: [{ id: 'fast', when: { kind: 'speed', speed: 'fast' }, inputPerMTok: 16, outputPerMTok: 80 }],
|
|
420
441
|
capability: 'frontier',
|
|
421
|
-
tier: 'frontier',
|
|
422
442
|
notes: 'Cache hits cost 2.5% of the input price ($0.20). Fast mode (usage.speed "fast") bills $16/$80.'
|
|
423
443
|
+ ' Requires 30-day data retention.',
|
|
424
444
|
},
|
|
@@ -445,7 +465,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
445
465
|
multipliers: { cacheRead: 0.025, cacheWrite5m: 1.25, cacheWrite1h: 2, batch: 0.5 },
|
|
446
466
|
tiers: [{ id: 'fast', when: { kind: 'speed', speed: 'fast' }, inputPerMTok: 16, outputPerMTok: 80 }],
|
|
447
467
|
capability: 'frontier',
|
|
448
|
-
tier: 'frontier',
|
|
449
468
|
notes:
|
|
450
469
|
'Available only to organizations verified through Anthropic\'s verification programs.'
|
|
451
470
|
+ ' Fast mode (usage.speed "fast") bills $16/$80. Requires 30-day data retention.',
|
|
@@ -473,7 +492,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
473
492
|
cacheMinTokens: 512,
|
|
474
493
|
multipliers: { cacheRead: 0.025, cacheWrite5m: 1.25, cacheWrite1h: 2, batch: 0.5 },
|
|
475
494
|
capability: 'frontier',
|
|
476
|
-
tier: 'frontier',
|
|
477
495
|
notes: 'Legacy: superseded by Claude Fable 5.2, which costs less ($8/$40). Cache hits cost 2.5% of the input'
|
|
478
496
|
+ ' price ($0.25). Requires 30-day data retention.',
|
|
479
497
|
},
|
|
@@ -495,7 +513,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
495
513
|
cacheMinTokens: 512,
|
|
496
514
|
multipliers: { cacheRead: 0.025, cacheWrite5m: 1.25, cacheWrite1h: 2, batch: 0.5 },
|
|
497
515
|
capability: 'frontier',
|
|
498
|
-
tier: 'frontier',
|
|
499
516
|
notes:
|
|
500
517
|
'Available only to organizations verified through Anthropic\'s verification programs.'
|
|
501
518
|
+ ' Requires 30-day data retention.',
|
|
@@ -511,7 +528,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
511
528
|
contextWindow: 1_000_000,
|
|
512
529
|
cacheMinTokens: 512,
|
|
513
530
|
capability: 'frontier',
|
|
514
|
-
tier: 'frontier',
|
|
515
531
|
notes: 'Legacy: superseded by Claude Fable 5.1 at the same prices. Requires 30-day data retention.',
|
|
516
532
|
},
|
|
517
533
|
{
|
|
@@ -526,7 +542,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
526
542
|
contextWindow: 1_000_000,
|
|
527
543
|
cacheMinTokens: 512,
|
|
528
544
|
capability: 'frontier',
|
|
529
|
-
tier: 'frontier',
|
|
530
545
|
notes:
|
|
531
546
|
'Available only to organizations verified through Anthropic\'s verification programs.'
|
|
532
547
|
+ ' Requires 30-day data retention.',
|
|
@@ -557,7 +572,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
557
572
|
*/
|
|
558
573
|
tiers: [{ id: 'fast', when: { kind: 'speed', speed: 'fast' }, inputPerMTok: 8, outputPerMTok: 40 }],
|
|
559
574
|
capability: 'large',
|
|
560
|
-
tier: 'opus',
|
|
561
575
|
notes: 'Cache hits cost 5% of the input price ($0.20), against 10% on Opus 5 and earlier.'
|
|
562
576
|
+ ' Fast mode (usage.speed "fast") bills $8/$40.',
|
|
563
577
|
},
|
|
@@ -574,7 +588,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
574
588
|
// Fast mode in research preview, $10/$50; same sources as Opus 5.5's.
|
|
575
589
|
tiers: [{ id: 'fast', when: { kind: 'speed', speed: 'fast' }, inputPerMTok: 10, outputPerMTok: 50 }],
|
|
576
590
|
capability: 'large',
|
|
577
|
-
tier: 'opus',
|
|
578
591
|
notes: '512-token cache minimum: caches prompts that would miss on Opus 4.6. Fast mode bills $10/$50.',
|
|
579
592
|
},
|
|
580
593
|
{
|
|
@@ -594,7 +607,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
594
607
|
*/
|
|
595
608
|
tiers: [{ id: 'fast', when: { kind: 'speed', speed: 'fast' }, inputPerMTok: 10, outputPerMTok: 50 }],
|
|
596
609
|
capability: 'large',
|
|
597
|
-
tier: 'opus',
|
|
598
610
|
notes: 'Fast mode bills $10/$50 until it is removed from this model on 2026-11-30.',
|
|
599
611
|
},
|
|
600
612
|
{
|
|
@@ -608,7 +620,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
608
620
|
contextWindow: 1_000_000,
|
|
609
621
|
cacheMinTokens: 2048,
|
|
610
622
|
capability: 'large',
|
|
611
|
-
tier: 'opus',
|
|
612
623
|
},
|
|
613
624
|
{
|
|
614
625
|
id: 'claude-opus-4-6',
|
|
@@ -621,7 +632,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
621
632
|
contextWindow: 1_000_000,
|
|
622
633
|
cacheMinTokens: 4096,
|
|
623
634
|
capability: 'large',
|
|
624
|
-
tier: 'opus',
|
|
625
635
|
},
|
|
626
636
|
{
|
|
627
637
|
/*
|
|
@@ -642,7 +652,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
642
652
|
contextWindow: 200_000,
|
|
643
653
|
cacheMinTokens: 4096,
|
|
644
654
|
capability: 'large',
|
|
645
|
-
tier: 'opus',
|
|
646
655
|
notes: '200K context window and a 64K output cap.',
|
|
647
656
|
},
|
|
648
657
|
{
|
|
@@ -668,7 +677,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
668
677
|
contextWindow: 200_000,
|
|
669
678
|
cacheMinTokens: 1024,
|
|
670
679
|
capability: 'large',
|
|
671
|
-
tier: 'opus',
|
|
672
680
|
notes:
|
|
673
681
|
'Retired on the Claude API on 2026-08-05; still served on Google Cloud (Google managed) and Bedrock.'
|
|
674
682
|
+ ' Anthropic names Claude Opus 4.8 as the replacement.',
|
|
@@ -689,7 +697,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
689
697
|
contextWindow: 200_000,
|
|
690
698
|
cacheMinTokens: 1024,
|
|
691
699
|
capability: 'large',
|
|
692
|
-
tier: 'opus',
|
|
693
700
|
notes:
|
|
694
701
|
'Retired on the Claude API on 2026-06-15; still served on Google Cloud (Google managed).'
|
|
695
702
|
+ ' Anthropic names Claude Opus 4.8 as the replacement.',
|
|
@@ -711,7 +718,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
711
718
|
cacheMinTokens: 512,
|
|
712
719
|
multipliers: { cacheRead: 0.05, cacheWrite5m: 1.25, cacheWrite1h: 2, batch: 0.5 },
|
|
713
720
|
capability: 'mid',
|
|
714
|
-
tier: 'sonnet',
|
|
715
721
|
notes: 'Cache hits cost 5% of the input price ($0.10), against 10% ($0.20) on Sonnet 5.',
|
|
716
722
|
},
|
|
717
723
|
{
|
|
@@ -725,7 +731,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
725
731
|
contextWindow: 1_000_000,
|
|
726
732
|
cacheMinTokens: 1024,
|
|
727
733
|
capability: 'mid',
|
|
728
|
-
tier: 'sonnet',
|
|
729
734
|
notes:
|
|
730
735
|
'Launched with 2/10 as introductory pricing through 2026-08-31; the scheduled'
|
|
731
736
|
+ ' increase to 3/15 on 2026-09-01 was cancelled and 2/10 is the standard price.',
|
|
@@ -741,7 +746,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
741
746
|
contextWindow: 1_000_000,
|
|
742
747
|
cacheMinTokens: 1024,
|
|
743
748
|
capability: 'mid',
|
|
744
|
-
tier: 'sonnet',
|
|
745
749
|
},
|
|
746
750
|
{
|
|
747
751
|
/*
|
|
@@ -764,7 +768,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
764
768
|
contextWindow: 200_000,
|
|
765
769
|
cacheMinTokens: 1024,
|
|
766
770
|
capability: 'mid',
|
|
767
|
-
tier: 'sonnet',
|
|
768
771
|
notes: 'Deprecated; retires 2026-11-30. Recommended replacement: Claude Sonnet 5.5.',
|
|
769
772
|
},
|
|
770
773
|
{
|
|
@@ -787,7 +790,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
787
790
|
contextWindow: 200_000,
|
|
788
791
|
cacheMinTokens: 1024,
|
|
789
792
|
capability: 'mid',
|
|
790
|
-
tier: 'sonnet',
|
|
791
793
|
notes:
|
|
792
794
|
'Retired on the Claude API on 2026-06-15; still served on Google Cloud (Google managed) and Bedrock.'
|
|
793
795
|
+ ' Anthropic names Claude Sonnet 5.5 as the replacement.',
|
|
@@ -828,7 +830,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
828
830
|
},
|
|
829
831
|
],
|
|
830
832
|
capability: 'small',
|
|
831
|
-
tier: 'haiku',
|
|
832
833
|
notes:
|
|
833
834
|
'Priced per request by prompt length: $0.10/$0.50 up to 100,000 input tokens, cache reads'
|
|
834
835
|
+ ' and writes included, and $0.50/$2.50 over it.',
|
|
@@ -847,7 +848,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
847
848
|
contextWindow: 200_000,
|
|
848
849
|
cacheMinTokens: 4096,
|
|
849
850
|
capability: 'small',
|
|
850
|
-
tier: 'haiku',
|
|
851
851
|
notes: '200K context window and a 64K output cap, smaller than the rest.',
|
|
852
852
|
},
|
|
853
853
|
{
|
|
@@ -869,7 +869,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
869
869
|
contextWindow: 200_000,
|
|
870
870
|
cacheMinTokens: 2048,
|
|
871
871
|
capability: 'small',
|
|
872
|
-
tier: 'haiku',
|
|
873
872
|
notes:
|
|
874
873
|
'Retired on the Claude API on 2026-02-19; still served on Google Cloud (Google managed).'
|
|
875
874
|
+ ' Anthropic names Claude Haiku 4.5 as the replacement.',
|
|
@@ -940,7 +939,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
940
939
|
ultrafast: { inputPerMTok: 60, outputPerMTok: 300, tiers: openAiLongBand(120, 450) },
|
|
941
940
|
},
|
|
942
941
|
capability: 'frontier',
|
|
943
|
-
tier: 'frontier',
|
|
944
942
|
notes: OPENAI_CURRENT_NOTE,
|
|
945
943
|
},
|
|
946
944
|
{
|
|
@@ -968,7 +966,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
968
966
|
ultrafast: { inputPerMTok: 12, outputPerMTok: 60, tiers: openAiLongBand(24, 90) },
|
|
969
967
|
},
|
|
970
968
|
capability: 'large',
|
|
971
|
-
tier: 'opus',
|
|
972
969
|
notes: OPENAI_CURRENT_NOTE,
|
|
973
970
|
},
|
|
974
971
|
{
|
|
@@ -1001,7 +998,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1001
998
|
fast: { inputPerMTok: 2, outputPerMTok: 10, tiers: openAiLongBand(4, 15) },
|
|
1002
999
|
},
|
|
1003
1000
|
capability: 'mid',
|
|
1004
|
-
tier: 'sonnet',
|
|
1005
1001
|
notes: OPENAI_CURRENT_NOTE,
|
|
1006
1002
|
},
|
|
1007
1003
|
{
|
|
@@ -1028,7 +1024,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1028
1024
|
fast: { inputPerMTok: 4, outputPerMTok: 20, tiers: openAiLongBand(8, 30) },
|
|
1029
1025
|
},
|
|
1030
1026
|
capability: 'large',
|
|
1031
|
-
tier: 'opus',
|
|
1032
1027
|
notes: OPENAI_CURRENT_NOTE,
|
|
1033
1028
|
},
|
|
1034
1029
|
{
|
|
@@ -1055,7 +1050,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1055
1050
|
fast: { inputPerMTok: 0.2, outputPerMTok: 1, tiers: openAiLongBand(0.4, 1.5) },
|
|
1056
1051
|
},
|
|
1057
1052
|
capability: 'small',
|
|
1058
|
-
tier: 'haiku',
|
|
1059
1053
|
notes: OPENAI_CURRENT_NOTE,
|
|
1060
1054
|
},
|
|
1061
1055
|
{
|
|
@@ -1096,7 +1090,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1096
1090
|
fast: { inputPerMTok: 8, outputPerMTok: 40, tiers: openAiLongBand(16, 60) },
|
|
1097
1091
|
},
|
|
1098
1092
|
capability: 'large',
|
|
1099
|
-
tier: 'opus',
|
|
1100
1093
|
notes:
|
|
1101
1094
|
OPENAI_CURRENT_NOTE
|
|
1102
1095
|
+ ' $4/$20 is promotional until 2026-11-30, and OpenAI has not published the price after it as a figure.',
|
|
@@ -1125,7 +1118,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1125
1118
|
fast: { inputPerMTok: 4, outputPerMTok: 24, tiers: openAiLongBand(8, 36) },
|
|
1126
1119
|
},
|
|
1127
1120
|
capability: 'mid',
|
|
1128
|
-
tier: 'sonnet',
|
|
1129
1121
|
notes: OPENAI_CURRENT_NOTE,
|
|
1130
1122
|
},
|
|
1131
1123
|
{
|
|
@@ -1152,7 +1144,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1152
1144
|
fast: { inputPerMTok: 0.4, outputPerMTok: 2.4, tiers: openAiLongBand(0.8, 3.6) },
|
|
1153
1145
|
},
|
|
1154
1146
|
capability: 'small',
|
|
1155
|
-
tier: 'haiku',
|
|
1156
1147
|
notes: OPENAI_CURRENT_NOTE,
|
|
1157
1148
|
},
|
|
1158
1149
|
{
|
|
@@ -1183,7 +1174,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1183
1174
|
multipliers: { cacheRead: 0.1, cacheWrite5m: 1.25, cacheWrite1h: 2.5, batch: null },
|
|
1184
1175
|
cacheTtl: OPENAI_CACHE_TTL,
|
|
1185
1176
|
capability: 'large',
|
|
1186
|
-
tier: 'opus',
|
|
1187
1177
|
notes:
|
|
1188
1178
|
'Needs approval through the Daybreak programme. At most 272,000 input tokens, Responses API only, no batch.'
|
|
1189
1179
|
+ ' Cache writes cost 1.25x input for the default 30-minute entry and 2.5x for a 12-hour one.',
|
|
@@ -1211,7 +1201,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1211
1201
|
fast: { inputPerMTok: 2.5, outputPerMTok: 20 },
|
|
1212
1202
|
},
|
|
1213
1203
|
capability: 'frontier',
|
|
1214
|
-
tier: 'frontier',
|
|
1215
1204
|
notes: OPENAI_GPT5_NOTE + OPENAI_DEPRECATED_SUFFIX,
|
|
1216
1205
|
},
|
|
1217
1206
|
{
|
|
@@ -1231,7 +1220,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1231
1220
|
fast: { inputPerMTok: 0.45, outputPerMTok: 3.6 },
|
|
1232
1221
|
},
|
|
1233
1222
|
capability: 'mid',
|
|
1234
|
-
tier: 'sonnet',
|
|
1235
1223
|
notes: OPENAI_GPT5_NOTE + OPENAI_DEPRECATED_SUFFIX,
|
|
1236
1224
|
},
|
|
1237
1225
|
{
|
|
@@ -1250,7 +1238,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1250
1238
|
flex: { inputPerMTok: 0.025, outputPerMTok: 0.2 },
|
|
1251
1239
|
},
|
|
1252
1240
|
capability: 'small',
|
|
1253
|
-
tier: 'haiku',
|
|
1254
1241
|
notes: OPENAI_GPT5_NOTE + OPENAI_DEPRECATED_SUFFIX,
|
|
1255
1242
|
},
|
|
1256
1243
|
|
|
@@ -1294,7 +1281,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1294
1281
|
fast: { inputPerMTok: 12.5, outputPerMTok: 75 },
|
|
1295
1282
|
},
|
|
1296
1283
|
capability: 'frontier',
|
|
1297
|
-
tier: 'frontier',
|
|
1298
1284
|
notes: OPENAI_EARLIER_LONG_NOTE,
|
|
1299
1285
|
},
|
|
1300
1286
|
{
|
|
@@ -1320,7 +1306,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1320
1306
|
flex: { inputPerMTok: 15, outputPerMTok: 90 },
|
|
1321
1307
|
},
|
|
1322
1308
|
capability: 'frontier',
|
|
1323
|
-
tier: 'frontier',
|
|
1324
1309
|
notes: OPENAI_EARLIER_LONG_NOTE,
|
|
1325
1310
|
},
|
|
1326
1311
|
{
|
|
@@ -1347,7 +1332,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1347
1332
|
fast: { inputPerMTok: 5, outputPerMTok: 30 },
|
|
1348
1333
|
},
|
|
1349
1334
|
capability: 'frontier',
|
|
1350
|
-
tier: 'frontier',
|
|
1351
1335
|
notes: OPENAI_EARLIER_LONG_NOTE,
|
|
1352
1336
|
},
|
|
1353
1337
|
{
|
|
@@ -1373,7 +1357,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1373
1357
|
flex: { inputPerMTok: 15, outputPerMTok: 90, tiers: openAiLongBand(30, 135) },
|
|
1374
1358
|
},
|
|
1375
1359
|
capability: 'frontier',
|
|
1376
|
-
tier: 'frontier',
|
|
1377
1360
|
notes: OPENAI_EARLIER_LONG_NOTE,
|
|
1378
1361
|
},
|
|
1379
1362
|
{
|
|
@@ -1392,7 +1375,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1392
1375
|
fast: { inputPerMTok: 1.5, outputPerMTok: 9 },
|
|
1393
1376
|
},
|
|
1394
1377
|
capability: 'mid',
|
|
1395
|
-
tier: 'sonnet',
|
|
1396
1378
|
notes: OPENAI_EARLIER_NOTE,
|
|
1397
1379
|
},
|
|
1398
1380
|
{
|
|
@@ -1411,7 +1393,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1411
1393
|
flex: { inputPerMTok: 0.1, outputPerMTok: 0.625 },
|
|
1412
1394
|
},
|
|
1413
1395
|
capability: 'small',
|
|
1414
|
-
tier: 'haiku',
|
|
1415
1396
|
notes: OPENAI_EARLIER_NOTE + OPENAI_DEPRECATED_SUFFIX,
|
|
1416
1397
|
},
|
|
1417
1398
|
{
|
|
@@ -1430,7 +1411,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1430
1411
|
fast: { inputPerMTok: 3.5, outputPerMTok: 28 },
|
|
1431
1412
|
},
|
|
1432
1413
|
capability: 'frontier',
|
|
1433
|
-
tier: 'frontier',
|
|
1434
1414
|
notes: OPENAI_EARLIER_NOTE,
|
|
1435
1415
|
},
|
|
1436
1416
|
{
|
|
@@ -1445,7 +1425,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1445
1425
|
caching: 'none',
|
|
1446
1426
|
multipliers: { batch: 0.5 },
|
|
1447
1427
|
capability: 'frontier',
|
|
1448
|
-
tier: 'frontier',
|
|
1449
1428
|
notes: OPENAI_EARLIER_NO_CACHE_NOTE,
|
|
1450
1429
|
},
|
|
1451
1430
|
{
|
|
@@ -1465,7 +1444,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1465
1444
|
fast: { inputPerMTok: 2.5, outputPerMTok: 20 },
|
|
1466
1445
|
},
|
|
1467
1446
|
capability: 'frontier',
|
|
1468
|
-
tier: 'frontier',
|
|
1469
1447
|
notes: OPENAI_EARLIER_NOTE + OPENAI_DEPRECATED_SUFFIX,
|
|
1470
1448
|
},
|
|
1471
1449
|
{
|
|
@@ -1481,7 +1459,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1481
1459
|
caching: 'none',
|
|
1482
1460
|
multipliers: { batch: 0.5 },
|
|
1483
1461
|
capability: 'frontier',
|
|
1484
|
-
tier: 'frontier',
|
|
1485
1462
|
notes: OPENAI_EARLIER_NO_CACHE_NOTE + OPENAI_DEPRECATED_SUFFIX,
|
|
1486
1463
|
},
|
|
1487
1464
|
{
|
|
@@ -1500,7 +1477,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1500
1477
|
fast: { inputPerMTok: 3.5, outputPerMTok: 14 },
|
|
1501
1478
|
},
|
|
1502
1479
|
capability: 'large',
|
|
1503
|
-
tier: 'opus',
|
|
1504
1480
|
notes: OPENAI_EARLIER_NOTE + OPENAI_DEPRECATED_SUFFIX,
|
|
1505
1481
|
},
|
|
1506
1482
|
{
|
|
@@ -1519,7 +1495,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1519
1495
|
fast: { inputPerMTok: 0.7, outputPerMTok: 2.8 },
|
|
1520
1496
|
},
|
|
1521
1497
|
capability: 'mid',
|
|
1522
|
-
tier: 'sonnet',
|
|
1523
1498
|
notes: OPENAI_EARLIER_NOTE + OPENAI_DEPRECATED_SUFFIX,
|
|
1524
1499
|
},
|
|
1525
1500
|
{
|
|
@@ -1538,7 +1513,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1538
1513
|
fast: { inputPerMTok: 4.25, outputPerMTok: 17 },
|
|
1539
1514
|
},
|
|
1540
1515
|
capability: 'large',
|
|
1541
|
-
tier: 'opus',
|
|
1542
1516
|
notes: OPENAI_EARLIER_NOTE + OPENAI_DEPRECATED_SUFFIX,
|
|
1543
1517
|
},
|
|
1544
1518
|
{
|
|
@@ -1557,7 +1531,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1557
1531
|
fast: { inputPerMTok: 0.25, outputPerMTok: 1 },
|
|
1558
1532
|
},
|
|
1559
1533
|
capability: 'small',
|
|
1560
|
-
tier: 'haiku',
|
|
1561
1534
|
notes: OPENAI_EARLIER_NOTE + OPENAI_DEPRECATED_SUFFIX,
|
|
1562
1535
|
},
|
|
1563
1536
|
{
|
|
@@ -1577,7 +1550,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1577
1550
|
fast: { inputPerMTok: 3.5, outputPerMTok: 14 },
|
|
1578
1551
|
},
|
|
1579
1552
|
capability: 'frontier',
|
|
1580
|
-
tier: 'frontier',
|
|
1581
1553
|
notes: OPENAI_EARLIER_NOTE + OPENAI_DEPRECATED_SUFFIX,
|
|
1582
1554
|
},
|
|
1583
1555
|
{
|
|
@@ -1593,7 +1565,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1593
1565
|
caching: 'none',
|
|
1594
1566
|
multipliers: { batch: 0.5 },
|
|
1595
1567
|
capability: 'frontier',
|
|
1596
|
-
tier: 'frontier',
|
|
1597
1568
|
notes: OPENAI_EARLIER_NO_CACHE_NOTE + OPENAI_DEPRECATED_SUFFIX,
|
|
1598
1569
|
},
|
|
1599
1570
|
// ------------------------------------------------------------------------
|
|
@@ -1625,7 +1596,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1625
1596
|
// The pricing page's specialized "Fast" table: $3.50/$28, cached $0.35.
|
|
1626
1597
|
serviceTiers: { fast: { inputPerMTok: 3.5, outputPerMTok: 28 } },
|
|
1627
1598
|
capability: 'frontier',
|
|
1628
|
-
tier: 'frontier',
|
|
1629
1599
|
notes: OPENAI_CODEX_NOTE + ' At most 272,000 input tokens.' + OPENAI_DEPRECATED_SUFFIX,
|
|
1630
1600
|
},
|
|
1631
1601
|
{
|
|
@@ -1640,7 +1610,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1640
1610
|
caching: 'automatic',
|
|
1641
1611
|
multipliers: { cacheRead: 0.1, cacheWrite5m: 1, cacheWrite1h: 1, batch: null },
|
|
1642
1612
|
capability: 'frontier',
|
|
1643
|
-
tier: 'frontier',
|
|
1644
1613
|
notes: OPENAI_CODEX_NOTE + ' At most 272,000 input tokens.' + OPENAI_SHUT_DOWN_SUFFIX,
|
|
1645
1614
|
},
|
|
1646
1615
|
{
|
|
@@ -1655,7 +1624,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1655
1624
|
caching: 'automatic',
|
|
1656
1625
|
multipliers: { cacheRead: 0.1, cacheWrite5m: 1, cacheWrite1h: 1, batch: null },
|
|
1657
1626
|
capability: 'frontier',
|
|
1658
|
-
tier: 'frontier',
|
|
1659
1627
|
notes: OPENAI_CODEX_NOTE + OPENAI_SHUT_DOWN_SUFFIX,
|
|
1660
1628
|
},
|
|
1661
1629
|
{
|
|
@@ -1670,7 +1638,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1670
1638
|
caching: 'automatic',
|
|
1671
1639
|
multipliers: { cacheRead: 0.1, cacheWrite5m: 1, cacheWrite1h: 1, batch: null },
|
|
1672
1640
|
capability: 'frontier',
|
|
1673
|
-
tier: 'frontier',
|
|
1674
1641
|
notes: OPENAI_CODEX_NOTE + OPENAI_SHUT_DOWN_SUFFIX,
|
|
1675
1642
|
},
|
|
1676
1643
|
{
|
|
@@ -1685,7 +1652,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1685
1652
|
caching: 'automatic',
|
|
1686
1653
|
multipliers: { cacheRead: 0.1, cacheWrite5m: 1, cacheWrite1h: 1, batch: null },
|
|
1687
1654
|
capability: 'mid',
|
|
1688
|
-
tier: 'sonnet',
|
|
1689
1655
|
notes: OPENAI_CODEX_NOTE + OPENAI_SHUT_DOWN_SUFFIX,
|
|
1690
1656
|
},
|
|
1691
1657
|
{
|
|
@@ -1700,7 +1666,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1700
1666
|
caching: 'automatic',
|
|
1701
1667
|
multipliers: { cacheRead: 0.1, cacheWrite5m: 1, cacheWrite1h: 1, batch: null },
|
|
1702
1668
|
capability: 'frontier',
|
|
1703
|
-
tier: 'frontier',
|
|
1704
1669
|
notes: OPENAI_CODEX_NOTE + ' At most 272,000 input tokens.' + OPENAI_SHUT_DOWN_SUFFIX,
|
|
1705
1670
|
},
|
|
1706
1671
|
{
|
|
@@ -1715,7 +1680,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1715
1680
|
caching: 'automatic',
|
|
1716
1681
|
multipliers: { cacheRead: 0.25, cacheWrite5m: 1, cacheWrite1h: 1, batch: null },
|
|
1717
1682
|
capability: 'mid',
|
|
1718
|
-
tier: 'sonnet',
|
|
1719
1683
|
notes: OPENAI_CODEX_NOTE + ' A cache read costs 0.25x input.' + OPENAI_SHUT_DOWN_SUFFIX,
|
|
1720
1684
|
},
|
|
1721
1685
|
{
|
|
@@ -1730,7 +1694,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1730
1694
|
caching: 'automatic',
|
|
1731
1695
|
multipliers: { cacheRead: 0.1, cacheWrite5m: 1, cacheWrite1h: 1, batch: null },
|
|
1732
1696
|
capability: 'large',
|
|
1733
|
-
tier: 'opus',
|
|
1734
1697
|
notes: 'Points at the model ChatGPT currently uses, so it is priced and never recommended. Caching is automatic; a write costs what input costs. At most 272,000 input tokens, and no batch.',
|
|
1735
1698
|
},
|
|
1736
1699
|
|
|
@@ -1760,8 +1723,17 @@ export const MODELS: ModelPricing[] = [
|
|
|
1760
1723
|
outputPerMTok: 7.5,
|
|
1761
1724
|
promo: { inputPerMTok: 0.75, outputPerMTok: 3.75, until: '2026-12-31', confirmedOn: '2026-11-16' },
|
|
1762
1725
|
history: [
|
|
1763
|
-
/*
|
|
1764
|
-
|
|
1726
|
+
/*
|
|
1727
|
+
* Its own price: $0.125 cached (10%) and batch half. Its Batch tab charges
|
|
1728
|
+
* the cache at the standard $0.125 against a batch input of $0.625 (and
|
|
1729
|
+
* $0.25 against $1.25 above 200k): 0.2 of the batch rate, read 2026-11-17.
|
|
1730
|
+
*/
|
|
1731
|
+
{
|
|
1732
|
+
until: '2026-12-01T00:00:00Z',
|
|
1733
|
+
inputPerMTok: 1.25,
|
|
1734
|
+
outputPerMTok: 10,
|
|
1735
|
+
multipliers: { cacheRead: 0.1, cacheWrite5m: 1, cacheWrite1h: 1, batch: 0.5, batchCacheRead: 0.2 },
|
|
1736
|
+
},
|
|
1765
1737
|
],
|
|
1766
1738
|
contextWindow: 1_048_576,
|
|
1767
1739
|
cacheMinTokens: 2048,
|
|
@@ -1775,7 +1747,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1775
1747
|
'This model models/gemini-2.5-pro is no longer available to new users. Please update your code to use models/gemini-3.1-pro-preview for the latest features and improvements. We recommend you to use the Interactions API.',
|
|
1776
1748
|
},
|
|
1777
1749
|
capability: 'large',
|
|
1778
|
-
tier: 'opus',
|
|
1779
1750
|
notes:
|
|
1780
1751
|
'From 2026-12-01 Google routes this id to Gemini 3.8 Flash and bills it at 3.8 Flash rates.'
|
|
1781
1752
|
+ ' Before then it bills $1.25/$10, and $2.50/$15.00 for a prompt over 200k tokens; this row prices every prompt'
|
|
@@ -1797,6 +1768,21 @@ export const MODELS: ModelPricing[] = [
|
|
|
1797
1768
|
displayName: 'Gemini 2.5 Flash',
|
|
1798
1769
|
inputPerMTok: 0.3,
|
|
1799
1770
|
outputPerMTok: 2.5,
|
|
1771
|
+
history: [
|
|
1772
|
+
/*
|
|
1773
|
+
* The same $0.30/$2.50, and one figure that differs until the rerouting:
|
|
1774
|
+
* its own Batch tab charges the cache at the standard $0.03 against a
|
|
1775
|
+
* batch input of $0.15, 0.2 of the batch rate (read 2026-11-17). From
|
|
1776
|
+
* 2026-12-01 it is billed at 3.5 Flash-Lite's rates, whose Batch tab
|
|
1777
|
+
* prints $0.015.
|
|
1778
|
+
*/
|
|
1779
|
+
{
|
|
1780
|
+
until: '2026-12-01T00:00:00Z',
|
|
1781
|
+
inputPerMTok: 0.3,
|
|
1782
|
+
outputPerMTok: 2.5,
|
|
1783
|
+
multipliers: { cacheRead: 0.1, cacheWrite5m: 1, cacheWrite1h: 1, batch: 0.5, batchCacheRead: 0.2 },
|
|
1784
|
+
},
|
|
1785
|
+
],
|
|
1800
1786
|
contextWindow: 1_048_576,
|
|
1801
1787
|
cacheMinTokens: 2048,
|
|
1802
1788
|
caching: 'automatic',
|
|
@@ -1809,7 +1795,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1809
1795
|
'This model models/gemini-2.5-flash is no longer available to new users. Please update your code to use models/gemini-3.6-flash for the latest features and improvements. We recommend you to use the Interactions API.',
|
|
1810
1796
|
},
|
|
1811
1797
|
capability: 'mid',
|
|
1812
|
-
tier: 'sonnet',
|
|
1813
1798
|
notes:
|
|
1814
1799
|
'From 2026-12-01 Google routes this id to Gemini 3.5 Flash-Lite and bills it at 3.5 Flash-Lite rates,'
|
|
1815
1800
|
+ ' which for text are the $0.30/$2.50 and $0.03 cached this row carries.',
|
|
@@ -1853,7 +1838,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1853
1838
|
*/
|
|
1854
1839
|
multipliers: { cacheRead: 0.1, cacheWrite5m: 1, cacheWrite1h: 1, batch: 0.5 },
|
|
1855
1840
|
capability: 'mid',
|
|
1856
|
-
tier: 'sonnet',
|
|
1857
1841
|
notes:
|
|
1858
1842
|
'Implicit caching is automatic above 4,096 tokens. Explicit caching is also billed for storage:'
|
|
1859
1843
|
+ ' $0.50 per 1M tokens per hour through 2026-12-31, $1.00 after.',
|
|
@@ -1882,7 +1866,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1882
1866
|
/* $0.075 / $0.75 = 0.1 (and $0.15 / $1.50); batch $0.375 / $0.75 = 0.5 (and $1.875 / $3.75). */
|
|
1883
1867
|
multipliers: { cacheRead: 0.1, cacheWrite5m: 1, cacheWrite1h: 1, batch: 0.5 },
|
|
1884
1868
|
capability: 'mid',
|
|
1885
|
-
tier: 'sonnet',
|
|
1886
1869
|
notes:
|
|
1887
1870
|
'Implicit caching is automatic above 4,096 tokens. Explicit caching is also billed for storage:'
|
|
1888
1871
|
+ ' $0.50 per 1M tokens per hour through 2026-12-31, $1.00 after.',
|
|
@@ -1912,7 +1895,12 @@ export const MODELS: ModelPricing[] = [
|
|
|
1912
1895
|
contextWindow: 1_048_576,
|
|
1913
1896
|
cacheMinTokens: 4096,
|
|
1914
1897
|
caching: 'automatic',
|
|
1915
|
-
|
|
1898
|
+
/*
|
|
1899
|
+
* The Batch tab's cache row reads "Same as Standard": $0.20 against a batch
|
|
1900
|
+
* input of $1.00, and $0.40 against $2.00 above 200k. 0.2 of the batch rate
|
|
1901
|
+
* in both bands, where a standard call reads at 0.1 (read 2026-11-17).
|
|
1902
|
+
*/
|
|
1903
|
+
multipliers: { cacheRead: 0.1, cacheWrite5m: 1, cacheWrite1h: 1, batch: 0.5, batchCacheRead: 0.2 },
|
|
1916
1904
|
tiers: [
|
|
1917
1905
|
{
|
|
1918
1906
|
id: 'over-200k',
|
|
@@ -1922,7 +1910,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1922
1910
|
},
|
|
1923
1911
|
],
|
|
1924
1912
|
capability: 'large',
|
|
1925
|
-
tier: 'opus',
|
|
1926
1913
|
notes:
|
|
1927
1914
|
'Priced per request by prompt length: $2/$12 up to 200,000 input tokens and $4/$18 over it.'
|
|
1928
1915
|
+ ' Implicit caching is automatic above 4,096 tokens; explicit caching is also billed for storage,'
|
|
@@ -1968,7 +1955,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1968
1955
|
caching: 'automatic',
|
|
1969
1956
|
multipliers: { cacheRead: 0.05, cacheWrite5m: 1, cacheWrite1h: 1, batch: 0.5 },
|
|
1970
1957
|
capability: 'frontier',
|
|
1971
|
-
tier: 'frontier',
|
|
1972
1958
|
notes: GEMINI_4_NOTE,
|
|
1973
1959
|
},
|
|
1974
1960
|
{
|
|
@@ -1983,7 +1969,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1983
1969
|
caching: 'automatic',
|
|
1984
1970
|
multipliers: { cacheRead: 0.05, cacheWrite5m: 1, cacheWrite1h: 1, batch: 0.5 },
|
|
1985
1971
|
capability: 'frontier',
|
|
1986
|
-
tier: 'frontier',
|
|
1987
1972
|
notes: GEMINI_4_NOTE,
|
|
1988
1973
|
},
|
|
1989
1974
|
{
|
|
@@ -1997,7 +1982,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
1997
1982
|
caching: 'automatic',
|
|
1998
1983
|
multipliers: { cacheRead: 0.05, cacheWrite5m: 1, cacheWrite1h: 1, batch: 0.5 },
|
|
1999
1984
|
capability: 'mid',
|
|
2000
|
-
tier: 'sonnet',
|
|
2001
1985
|
notes: GEMINI_4_NOTE,
|
|
2002
1986
|
},
|
|
2003
1987
|
{
|
|
@@ -2022,7 +2006,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2022
2006
|
caching: 'automatic',
|
|
2023
2007
|
multipliers: { cacheRead: 0.1, cacheWrite5m: 1, cacheWrite1h: 1, batch: 0.5 },
|
|
2024
2008
|
capability: 'small',
|
|
2025
|
-
tier: 'haiku',
|
|
2026
2009
|
notes:
|
|
2027
2010
|
'Implicit caching is automatic, and no minimum prefix length is published for this model. Explicit caching is'
|
|
2028
2011
|
+ ' also billed for storage, $1.00 per 1M tokens per hour.',
|
|
@@ -2064,7 +2047,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2064
2047
|
caching: 'automatic',
|
|
2065
2048
|
multipliers: { cacheRead: 0.1, cacheWrite5m: 1, cacheWrite1h: 2, batch: 0.6 },
|
|
2066
2049
|
capability: 'frontier',
|
|
2067
|
-
tier: 'frontier',
|
|
2068
2050
|
notes:
|
|
2069
2051
|
'Cache writes are billed as their own line: 1x input for a 5-minute entry, 2x for an hour.'
|
|
2070
2052
|
+ ` ${NO_PUBLISHED_CACHE_MINIMUM}`,
|
|
@@ -2084,7 +2066,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2084
2066
|
caching: 'automatic',
|
|
2085
2067
|
multipliers: { cacheRead: 0.05, cacheWrite5m: 1, cacheWrite1h: 2, batch: null },
|
|
2086
2068
|
capability: 'frontier',
|
|
2087
|
-
tier: 'frontier',
|
|
2088
2069
|
notes:
|
|
2089
2070
|
'Cache writes are billed as their own line: 1x input for a 5-minute entry, 2x for an hour.'
|
|
2090
2071
|
+ ` No batch API for this model. ${NO_PUBLISHED_CACHE_MINIMUM}`,
|
|
@@ -2100,7 +2081,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2100
2081
|
caching: 'automatic',
|
|
2101
2082
|
multipliers: { cacheRead: 0.1, cacheWrite5m: 1, cacheWrite1h: 2, batch: 0.6 },
|
|
2102
2083
|
capability: 'frontier',
|
|
2103
|
-
tier: 'frontier',
|
|
2104
2084
|
notes:
|
|
2105
2085
|
'Cache writes are billed as their own line: 1x input for a 5-minute entry, 2x for an hour.'
|
|
2106
2086
|
+ ` Batch at 60% since the batch page listed it (read 2026-11-05). ${NO_PUBLISHED_CACHE_MINIMUM}`,
|
|
@@ -2119,7 +2099,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2119
2099
|
/* $0.19 on a hit against $0.95; batch $0.57 / $2.40, 60%. */
|
|
2120
2100
|
multipliers: { cacheRead: 0.2, cacheWrite5m: 1, cacheWrite1h: 1, batch: 0.6 },
|
|
2121
2101
|
capability: 'mid',
|
|
2122
|
-
tier: 'sonnet',
|
|
2123
2102
|
notes: NO_PUBLISHED_CACHE_MINIMUM,
|
|
2124
2103
|
},
|
|
2125
2104
|
{
|
|
@@ -2136,7 +2115,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2136
2115
|
/* $0.38 on a hit against $1.90; no batch API for this variant. */
|
|
2137
2116
|
multipliers: { cacheRead: 0.2, cacheWrite5m: 1, cacheWrite1h: 1, batch: null },
|
|
2138
2117
|
capability: 'mid',
|
|
2139
|
-
tier: 'sonnet',
|
|
2140
2118
|
notes: NO_PUBLISHED_CACHE_MINIMUM,
|
|
2141
2119
|
},
|
|
2142
2120
|
{
|
|
@@ -2153,7 +2131,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2153
2131
|
/* $0.16 on a hit against $0.95, written as the ratio so the product is the page's figure. */
|
|
2154
2132
|
multipliers: { cacheRead: 0.16 / 0.95, cacheWrite5m: 1, cacheWrite1h: 1, batch: 0.6 },
|
|
2155
2133
|
capability: 'mid',
|
|
2156
|
-
tier: 'sonnet',
|
|
2157
2134
|
notes: NO_PUBLISHED_CACHE_MINIMUM,
|
|
2158
2135
|
},
|
|
2159
2136
|
{
|
|
@@ -2176,7 +2153,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2176
2153
|
*/
|
|
2177
2154
|
recommendable: false,
|
|
2178
2155
|
capability: 'mid',
|
|
2179
|
-
tier: 'sonnet',
|
|
2180
2156
|
notes:
|
|
2181
2157
|
'Discontinued by Moonshot on 2026-05-25; priced at the last rate it published. No batch API.'
|
|
2182
2158
|
+ ` ${NO_PUBLISHED_CACHE_MINIMUM}`,
|
|
@@ -2202,7 +2178,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2202
2178
|
'The supported API model names are deepseek-v4-pro, deepseek-v4-flash, and deepseek-v4-flash-vision-exp, but you passed deepseek-v3.',
|
|
2203
2179
|
},
|
|
2204
2180
|
capability: 'mid',
|
|
2205
|
-
tier: 'sonnet',
|
|
2206
2181
|
},
|
|
2207
2182
|
{
|
|
2208
2183
|
/*
|
|
@@ -2259,7 +2234,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2259
2234
|
/* $0.003 cache hit against $0.15 input off-peak, $0.006 against $0.30 at peak. */
|
|
2260
2235
|
multipliers: { cacheRead: 0.02, cacheWrite5m: 1, cacheWrite1h: 1, batch: null },
|
|
2261
2236
|
capability: 'mid',
|
|
2262
|
-
tier: 'sonnet',
|
|
2263
2237
|
notes:
|
|
2264
2238
|
'Retired 2026-09-10: the id is still accepted, served by V4.1 Flash and billed at the deepseek-flash price.'
|
|
2265
2239
|
+ ' Maximum output 384K tokens. Peak excludes Chinese public holidays, which this table cannot express.',
|
|
@@ -2293,7 +2267,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2293
2267
|
caching: 'automatic',
|
|
2294
2268
|
multipliers: { cacheRead: 0.02, cacheWrite5m: 1, cacheWrite1h: 1, batch: null },
|
|
2295
2269
|
capability: 'mid',
|
|
2296
|
-
tier: 'sonnet',
|
|
2297
2270
|
notes:
|
|
2298
2271
|
'Maximum output 384K tokens. Peak excludes Chinese public holidays, which this table cannot express:'
|
|
2299
2272
|
+ ' on a weekday holiday it charges peak where DeepSeek charges off-peak.',
|
|
@@ -2319,7 +2292,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2319
2292
|
/* $0.022 cache hit against $0.66 input off-peak. */
|
|
2320
2293
|
multipliers: { cacheRead: 0.0333, cacheWrite5m: 1, cacheWrite1h: 1, batch: null },
|
|
2321
2294
|
capability: 'large',
|
|
2322
|
-
tier: 'opus',
|
|
2323
2295
|
notes: 'Maximum output 384K tokens.',
|
|
2324
2296
|
},
|
|
2325
2297
|
|
|
@@ -2348,7 +2320,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2348
2320
|
multipliers: { cacheRead: 0.25, cacheWrite5m: 1, cacheWrite1h: 1, batch: null },
|
|
2349
2321
|
tiers: [GROK_LONG(4, 12)],
|
|
2350
2322
|
capability: 'frontier',
|
|
2351
|
-
tier: 'frontier',
|
|
2352
2323
|
notes: NO_PUBLISHED_CACHE_MINIMUM,
|
|
2353
2324
|
},
|
|
2354
2325
|
{
|
|
@@ -2363,7 +2334,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2363
2334
|
multipliers: { cacheRead: 0.25, cacheWrite5m: 1, cacheWrite1h: 1, batch: null },
|
|
2364
2335
|
tiers: [GROK_LONG(4, 12)],
|
|
2365
2336
|
capability: 'large',
|
|
2366
|
-
tier: 'opus',
|
|
2367
2337
|
notes: NO_PUBLISHED_CACHE_MINIMUM,
|
|
2368
2338
|
},
|
|
2369
2339
|
{
|
|
@@ -2379,7 +2349,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2379
2349
|
multipliers: { cacheRead: 0.15, cacheWrite5m: 1, cacheWrite1h: 1, batch: null },
|
|
2380
2350
|
tiers: [GROK_LONG(4, 12)],
|
|
2381
2351
|
capability: 'large',
|
|
2382
|
-
tier: 'opus',
|
|
2383
2352
|
notes: NO_PUBLISHED_CACHE_MINIMUM,
|
|
2384
2353
|
},
|
|
2385
2354
|
{
|
|
@@ -2395,7 +2364,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2395
2364
|
multipliers: { cacheRead: 0.16, cacheWrite5m: 1, cacheWrite1h: 1, batch: 0.8 },
|
|
2396
2365
|
tiers: [GROK_LONG(2.5, 5)],
|
|
2397
2366
|
capability: 'mid',
|
|
2398
|
-
tier: 'sonnet',
|
|
2399
2367
|
notes: NO_PUBLISHED_CACHE_MINIMUM,
|
|
2400
2368
|
},
|
|
2401
2369
|
{
|
|
@@ -2411,7 +2379,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2411
2379
|
multipliers: { cacheRead: 0.16, cacheWrite5m: 1, cacheWrite1h: 1, batch: 0.8 },
|
|
2412
2380
|
tiers: [GROK_LONG(2.5, 5)],
|
|
2413
2381
|
capability: 'mid',
|
|
2414
|
-
tier: 'sonnet',
|
|
2415
2382
|
notes: NO_PUBLISHED_CACHE_MINIMUM,
|
|
2416
2383
|
},
|
|
2417
2384
|
{
|
|
@@ -2427,7 +2394,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2427
2394
|
multipliers: { cacheRead: 0.16, cacheWrite5m: 1, cacheWrite1h: 1, batch: 0.8 },
|
|
2428
2395
|
tiers: [GROK_LONG(2.5, 5)],
|
|
2429
2396
|
capability: 'mid',
|
|
2430
|
-
tier: 'sonnet',
|
|
2431
2397
|
notes: NO_PUBLISHED_CACHE_MINIMUM,
|
|
2432
2398
|
},
|
|
2433
2399
|
{
|
|
@@ -2443,7 +2409,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2443
2409
|
multipliers: { cacheRead: 0.16, cacheWrite5m: 1, cacheWrite1h: 1, batch: 0.8 },
|
|
2444
2410
|
tiers: [GROK_LONG(2.5, 5)],
|
|
2445
2411
|
capability: 'mid',
|
|
2446
|
-
tier: 'sonnet',
|
|
2447
2412
|
notes: NO_PUBLISHED_CACHE_MINIMUM,
|
|
2448
2413
|
},
|
|
2449
2414
|
{
|
|
@@ -2458,7 +2423,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2458
2423
|
multipliers: { cacheRead: 0.2, cacheWrite5m: 1, cacheWrite1h: 1, batch: null },
|
|
2459
2424
|
tiers: [GROK_LONG(2, 4)],
|
|
2460
2425
|
capability: 'mid',
|
|
2461
|
-
tier: 'sonnet',
|
|
2462
2426
|
notes: NO_PUBLISHED_CACHE_MINIMUM,
|
|
2463
2427
|
},
|
|
2464
2428
|
{
|
|
@@ -2491,7 +2455,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2491
2455
|
],
|
|
2492
2456
|
recommendable: false,
|
|
2493
2457
|
capability: 'large',
|
|
2494
|
-
tier: 'opus',
|
|
2495
2458
|
notes:
|
|
2496
2459
|
'Retired by xAI on 2026-05-15 and redirected to Grok 4.3, whose price it bills at from then;'
|
|
2497
2460
|
+ ` $3/$15 before. ${NO_PUBLISHED_CACHE_MINIMUM}`,
|
|
@@ -2517,7 +2480,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2517
2480
|
'Invalid model: mistral-large-2',
|
|
2518
2481
|
},
|
|
2519
2482
|
capability: 'large',
|
|
2520
|
-
tier: 'opus',
|
|
2521
2483
|
notes: 'No prompt caching: reordering still helps readability but saves nothing here.',
|
|
2522
2484
|
},
|
|
2523
2485
|
{
|
|
@@ -2548,7 +2510,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2548
2510
|
caching: 'unknown',
|
|
2549
2511
|
multipliers: { cacheRead: 0.1, cacheWrite5m: 1, cacheWrite1h: 1, batch: 0.5 },
|
|
2550
2512
|
capability: 'large',
|
|
2551
|
-
tier: 'opus',
|
|
2552
2513
|
notes: MISTRAL_CACHE_NOTE,
|
|
2553
2514
|
},
|
|
2554
2515
|
{
|
|
@@ -2576,7 +2537,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2576
2537
|
caching: 'unknown',
|
|
2577
2538
|
multipliers: { cacheRead: 0.1, cacheWrite5m: 1, cacheWrite1h: 1, batch: 0.5 },
|
|
2578
2539
|
capability: 'mid',
|
|
2579
|
-
tier: 'sonnet',
|
|
2580
2540
|
notes: MISTRAL_CACHE_NOTE,
|
|
2581
2541
|
},
|
|
2582
2542
|
{
|
|
@@ -2592,7 +2552,6 @@ export const MODELS: ModelPricing[] = [
|
|
|
2592
2552
|
caching: 'unknown',
|
|
2593
2553
|
multipliers: { cacheRead: 0.1, cacheWrite5m: 1, cacheWrite1h: 1, batch: 0.5 },
|
|
2594
2554
|
capability: 'small',
|
|
2595
|
-
tier: 'haiku',
|
|
2596
2555
|
notes: MISTRAL_CACHE_NOTE,
|
|
2597
2556
|
},
|
|
2598
2557
|
];
|
|
@@ -2695,14 +2654,17 @@ export function modelFrom(catalogue: PricingCatalogue, id: string): ModelPricing
|
|
|
2695
2654
|
}
|
|
2696
2655
|
|
|
2697
2656
|
/**
|
|
2698
|
-
* Cheapest model of a
|
|
2657
|
+
* Cheapest model of a capability within a catalogue.
|
|
2658
|
+
*
|
|
2659
|
+
* Takes the names `recommendTier` answers in, which are the names of the
|
|
2660
|
+
* `capability` a model carries.
|
|
2699
2661
|
*
|
|
2700
2662
|
* Compared on the **effective** price, so a model in a promotional window is
|
|
2701
2663
|
* ranked at what it actually costs today rather than at its list price.
|
|
2702
2664
|
*/
|
|
2703
|
-
export function
|
|
2665
|
+
export function cheapestOfCapabilityIn(
|
|
2704
2666
|
catalogue: PricingCatalogue,
|
|
2705
|
-
|
|
2667
|
+
capability: Capability,
|
|
2706
2668
|
on: Date = new Date(),
|
|
2707
2669
|
/**
|
|
2708
2670
|
* Restrict to one provider. Omit to search the whole catalogue, which is what
|
|
@@ -2714,12 +2676,12 @@ export function cheapestOfTierIn(
|
|
|
2714
2676
|
): ModelPricing {
|
|
2715
2677
|
const candidates = catalogue.models.filter(
|
|
2716
2678
|
(m) =>
|
|
2717
|
-
m.
|
|
2679
|
+
m.capability === capability &&
|
|
2718
2680
|
isOffered(m) &&
|
|
2719
2681
|
(provider === undefined || m.provider === provider),
|
|
2720
2682
|
);
|
|
2721
2683
|
const first = candidates[0];
|
|
2722
|
-
if (!first) throw new Error(`No models for
|
|
2684
|
+
if (!first) throw new Error(`No models for capability "${capability}"`);
|
|
2723
2685
|
return candidates.reduce(
|
|
2724
2686
|
(best, m) =>
|
|
2725
2687
|
effectivePricing(m, on).inputPerMTok < effectivePricing(best, on).inputPerMTok ? m : best,
|
|
@@ -3208,11 +3170,11 @@ export function reviewedForModels(
|
|
|
3208
3170
|
return oldest ?? catalogue.lastReviewed;
|
|
3209
3171
|
}
|
|
3210
3172
|
|
|
3211
|
-
/** Cheapest model of
|
|
3212
|
-
export function
|
|
3213
|
-
const candidates = MODELS.filter((m) => m.
|
|
3173
|
+
/** Cheapest model of a capability across the bundled catalogue, for recommendations. */
|
|
3174
|
+
export function cheapestOfCapability(capability: Capability): ModelPricing {
|
|
3175
|
+
const candidates = MODELS.filter((m) => m.capability === capability && isOffered(m));
|
|
3214
3176
|
const first = candidates[0];
|
|
3215
|
-
if (!first) throw new Error(`No models for
|
|
3177
|
+
if (!first) throw new Error(`No models for capability "${capability}"`);
|
|
3216
3178
|
return candidates.reduce((best, m) => (m.inputPerMTok < best.inputPerMTok ? m : best), first);
|
|
3217
3179
|
}
|
|
3218
3180
|
|