Research · October 2026

AI API Pricing Index — October 2026

Name: AI API Pricing Index — October 2026 frozen snapshot
Creator: AICostBudget
License: https://creativecommons.org/licenses/by/4.0/

A reproducible October month-end snapshot of 123 public pricing records across 13 providers. Every statistic below is calculated from the frozen public projection captured at the 2026-10-31 UTC cutoff. The live Dataset may contain newer records.

Cutoff 2026-10-31 UTCLast verified Oct 31, 2026 How we verifyCurrency USD

Explore the live dataset Read findings

Public records123

Frozen public projection

Providers13

First-party and cloud APIs

Pricing components999

Conditional structured charges

Token-priced records90

At least one per-1M-token component

Cache-priced records70

Comparable cached-input scalar

October events32

Verified, not inferred

Key findings

The useful signal is in the pricing structure, not one cheapest-model ranking.

All medians use frozen public records with a compatible numeric value. Missing values are excluded, never treated as zero.

$1.285

Median input price

Across 82 records with a current input-token scalar; observed range $0.05–$30 per 1M tokens.

$5.5

Median output price

Across 82 records with a current output-token scalar; observed range $0.10–$180 per 1M tokens.

5.0×

Median output/input ratio

Across 82 records with both prices. Output generation, not prompt ingestion, is commonly the higher list-price component.

90%

Median cache discount

Across 70 comparable records. The observed discount range is 75%–98%.

50%

Median Batch discount

Across 50 records with a calculation-default Batch input component; observed range 20%–50%.

Records with non-token charges

Document pages, storage, session duration, or time-based units cannot be safely flattened into a token-only price table.

Chart 1

Which providers contribute the most public records?

OpenAI25 records

Google Gemini20 records

Mistral AI16 records

Anthropic14 records

xAI14 records

Google Cloud6 records

AWS5 records

Cohere5 records

Moonshot AI5 records

Alibaba Qwen4 records

DeepSeek4 records

Azure3 records

Meta2 records

Source: frozen October 31, 2026 UTC public projection of the AICostBudget AI API Pricing Dataset. Methodology.

Chart 2

How do provider median input and output prices compare?

OpenAIInput $2Output $10n=19

Google GeminiInput $0.75Output $4.5n=16

Mistral AIInput $0.25Output $0.75n=8

AnthropicInput $4.5Output $22.5n=14

xAIInput $1.25Output $2.5n=9

CohereInput $1.5Output $5.75n=2

Moonshot AIInput $2Output $10n=5

Alibaba QwenInput $0.275Output $1.035n=4

DeepSeekInput $0.30Output $1.2n=3

MetaInput $2.625Output $12.125n=2

Source: frozen October 31, 2026 UTC public projection of the AICostBudget AI API Pricing Dataset. Methodology.

Chart 3

How widely are cost-saving price modes published?

70Cached-input recordsMedian discount 90%

50Batch-priced recordsMedian discount 50%

A workload comparison that ignores cache and Batch pricing omits published savings modes for a large share of the tracked records.

Source: frozen October 31, 2026 UTC public projection of the AICostBudget AI API Pricing Dataset. Methodology.

Chart 4

Where does token-only comparison stop working?

Token-priced90 records

Tiered / long-context24 records

Non-token components32 records

Legacy status3 records

Retired status7 records

Source: frozen October 31, 2026 UTC public projection of the AICostBudget AI API Pricing Dataset. Methodology.

Month over month

What changed since September 2026?

Frozen snapshots are compared directly; missing values remain excluded rather than treated as zero.

-3%

Median input price

$1.32 → $1.285

-8%

Median output price

$6 → $5.5

Median output/input ratio

5.0× → 5.0×

0 pp

Median cache discount

90% → 90%

0 pp

Median Batch discount

50% → 50%

42 / 3

Models added / retired

Records present now but absent from the previous snapshot (launches plus newly covered providers and pricing types), and records that moved to retired status.

Provider	Model	Change	Input	Output	Source
Mistral AI	Mistral Large 4	price update	$1.36 → $0.70 (-49%)	$4.18 → $2.1 (-50%)	Official source
Mistral AI	Mistral Large 4	price update	$0.68 → $1.36 (+100%)	$2.09 → $4.18 (+100%)	Official source

Most explicit changes: Mistral AI (2).

Chart 5 · Verified timeline

What changed in October 2026?

32 events: 13 model additions, 0 successor comparisons, 14 lifecycle updates, and 0 time-pricing schedule update. It also includes 5 price updates to list, cached, or component prices.

Oct 31

Mistral AI

Mistral Large 4

Token prices updated

Official source

Oct 29

Moonshot AI

Kimi K2.6

Model lifecycle updated

Official source

Oct 29

Moonshot AI

Kimi K2.7 Code

Model lifecycle updated

Official source

Oct 28

Model lifecycle updated

Official source

Source: frozen October 31, 2026 UTC public projection of the AICostBudget AI API Pricing Dataset. Methodology.

Crawlable data

Key statistics

The table states the population and method so each result can be independently reproduced.

Statistic	Result	Population	Method
Frozen public records	123	All frozen projection rows	Count records
Median input price	$1.285	82 non-null input scalars	Median USD per 1M tokens
Median output price	$5.5	82 non-null output scalars	Median USD per 1M tokens
Median output/input ratio	5.0×	82 rows with both scalars	Median of output ÷ input per record
Median cached-input discount	90%	70 comparable rows	Median of 1 − cached input ÷ input
Median Batch input discount	50%	50 Batch-capable rows	Median of 1 − Batch input ÷ standard input
Structured pricing components	999	All public projection rows	Count pricing_components entries
October verified events	32	Events effective 2026-10-01 to 2026-10-31	Count typed events; categories remain separate

Provider comparison

Coverage and frozen medians by provider

Provider medians describe the records tracked here; they are not market share or usage-weighted prices.

Provider	Records	Priced input n	Median input	Median output	Cache rows	Batch rows
OpenAI	25	19	$2	$10	14	13
Google Gemini	20	16	$0.75	$4.5	11	9
Mistral AI	16	8	$0.25	$0.75	8	7
Anthropic	14	14	$4.5	$22.5	14	11
xAI	14	9	$1.25	$2.5	9	8
Google Cloud	6	0	—	—	0	0
AWS	5	0	—	—	0	0
Cohere	5	2	$1.5	$5.75	0	0
Moonshot AI	5	5	$2	$10	5	2
Alibaba Qwen	4	4	$0.275	$1.035	4	0
DeepSeek	4	3	$0.30	$1.2	3	0
Azure	3	0	—	—	0	0
Meta	2	2	$2.625	$12.125	2	0

Methodology

What this index measures—and what it does not.

Source and scope

All statistics are calculated from the frozen AICostBudget public projection at the 2026-10-31 UTC cutoff. The population is 123 public pricing records. The live Dataset and downloads may contain newer records; later live updates do not silently change this historical Research page.

Normalization

Comparable text-token scalars are normalized to USD per 1M tokens. Provider and model aliases do not create a second research observation. Null means unavailable or not applicable, never zero.

Cache, Batch, and tiers

Cached input and Batch discounts are calculated only when both the standard scalar and a compatible published component exist. Long-context and usage tiers remain separate structured records; thresholds price the whole matching request.

Non-token pricing

Document pages, per-minute sessions, and token-storage hours keep their native billing units. They count as non-token pricing architecture and are excluded from input/output token medians.

Time and history

October events use provider-announced or official-changelog effective dates when available. The available event history does not support an industry-wide annual price-decline claim.

Verification limits

Public list prices are planning inputs, not invoices. Regional pricing, negotiated contracts, taxes, routing channels, tokenization, and workload quality can change realized cost.

Full field definitions, official source groups, exclusions, licensing, and downloads are maintained on the AI API Pricing Dataset methodology page.

Cite this research

Copy a citation or use the underlying data.

Link to this dated analysis for the findings; cite the Dataset when reusing the machine-readable records.

Plain text

AICostBudget. "AI API Pricing Index — October 2026." AICostBudget, 2026. Frozen public projection at the October 31, 2026 UTC cutoff. https://aicostbudget.com/en/research/ai-api-pricing-index-october-2026

BibTeX

@article{aicostbudget_pricing_index_2026_10,
  author = {{AICostBudget}},
  title = {AI API Pricing Index --- October 2026},
  year = {2026},
  month = {October},
  url = {https://aicostbudget.com/en/research/ai-api-pricing-index-october-2026},
  note = {Based on the AICostBudget AI API Pricing Dataset frozen public projection; cutoff 2026-10-31 UTC}
}

Monthly price changes CSV Live Dataset page JSON CSV GitHub Hugging Face

Previous editions

Each edition remains tied to its own frozen cutoff and source commit.

All research editions September 2026

Use the evidence

Move from list price to workload cost.

Current list prices answer only the first question. Apply them to your own context, output, cache, retry, and routing assumptions before making a model decision.

Track verified model price changesSee dated events and official sources.Compare current model pricesFilter the live comparison table.Estimate your API workloadApply token volume and request assumptions.Compare switching impactModel the before-and-after workload cost.Model retry overheadMeasure failed attempts and fallbacks.