# LLM API Pricing Comparison

> Source: https://aiwiki.ai/wiki/llm_api_pricing_comparison
> Updated: 2026-08-01
> Fact-checked: 2026-08-04
> Categories: AI Tools & Products, Large Language Models, Model Evaluation
> License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/) - attribute to "AI Wiki (aiwiki.ai)"
> Cite as: AI Wiki. "LLM API Pricing Comparison." aiwiki.ai, 1 Aug 2026. https://aiwiki.ai/wiki/llm_api_pricing_comparison
> From AI Wiki (https://aiwiki.ai), the free encyclopedia of artificial intelligence. Reuse freely with attribution.

Every figure in this article is a US dollar list rate per 1,000,000 tokens, taken from the provider's own pricing page and checked on **August 1, 2026**. That date matters more than any single number below. [OpenAI](https://aiwiki.ai/wiki/openai) cut the list price of [GPT-5.6](https://aiwiki.ai/wiki/gpt_5_6) Luna by 80 percent on July 30, two days before this revision, and [Anthropic](https://aiwiki.ai/wiki/anthropic) has a scheduled 50 percent increase on [Claude Sonnet 5](https://aiwiki.ai/wiki/claude_sonnet_5) landing on September 1. Prices in this market change without notice, sometimes twice in a quarter, so treat the tables as a snapshot and re-check the linked source pages before you commit a budget.

The short answer as of that date: the cheapest capable API is [DeepSeek V4-Flash](https://aiwiki.ai/wiki/deepseek_v4_flash) at $0.14 input and $0.28 output, with a cache-hit input rate of $0.0028 that is the deepest cache discount any major provider publishes.[8] The cheapest frontier-class model is [DeepSeek V4-Pro](https://aiwiki.ai/wiki/deepseek_v4) at $0.435 and $0.87, and both ship as [open weights](https://aiwiki.ai/wiki/open_weights) under the MIT license, so the API rate is only one way to buy them.[8][28] Among the largest US labs, OpenAI's GPT-5.6 Luna at $0.20 and $1.20 is the cheapest model that still scores near the top of independent aggregate rankings, while the premium flagships cluster at $5 input: GPT-5.6 Sol at $5 and $30, [Claude Opus 5](https://aiwiki.ai/wiki/claude_opus_5) at $5 and $25.[1][6] [Google](https://aiwiki.ai/wiki/google)'s [Gemini 3.1 Pro](https://aiwiki.ai/wiki/gemini_3_1_pro) undercuts both at $2 and $12 for prompts up to 200,000 tokens.[7]

That is where the naive comparison ends and the useful one begins. A per-token rate is now a weak predictor of what a job costs, because two models at similar rates can emit output token counts that differ by two orders of magnitude for the same task. The sections after the tables cover the things that actually move a bill: reasoning effort, cache state, prompt-length thresholds, service tiers, and where you buy the model.

This is the developer, per-token API pricing hub. For AI subscriptions, consumer plans, and market pricing generally, see the [AI pricing](https://aiwiki.ai/wiki/ai_pricing) concept page.

## Scope, units, and how volatile this is

Unless a row says otherwise, every price is:

- a public list rate, not a negotiated enterprise rate;
- in US dollars per 1,000,000 tokens;
- on the standard synchronous tier (batch, flex, and priority tiers are covered separately);
- on the provider's own first-party API, not a reseller or cloud marketplace;
- at the short-context tier, where a provider charges more above a token threshold.

Where a provider does not publish a distinct cache-hit rate, the tables say "not published" rather than assuming one. Where a provider's current price could not be confirmed on an official page, that provider is left out entirely rather than estimated. Several vendors that appeared in earlier revisions of this article, including xAI's Grok 4.1 Fast, are no longer listed on the vendor's own model page and have been removed for that reason.

Recent and scheduled changes to the rates below:

| Date | Change | Provider |
| --- | --- | --- |
| July 30, 2026 | GPT-5.6 Luna list price cut 80 percent; GPT-5.6 Terra cut 20 percent[2] | OpenAI |
| July 30, 2026 | "Priority processing" renamed "Fast mode"; `service_tier: "priority"` still accepted[1][30] | OpenAI |
| September 1, 2026 (scheduled) | Claude Sonnet 5 introductory $2 / $10 reverts to $3 / $15[6] | Anthropic |
| Announced, no effective date | Peak-hour pricing at 2x list, 9:00-12:00 and 14:00-18:00 Beijing time[8] | DeepSeek |

## What API pricing actually varies on

Nine things move a bill. Only the first is a headline number.

| Axis | What it is | Size of the effect |
| --- | --- | --- |
| Direction of the token | Output is billed above input everywhere | 2x to about 8x across the models below |
| Cache state | Cache miss, cache write, and cache hit are three different prices | Hits are 90 percent off at most providers and about 98 percent off at DeepSeek; writes cost 1.25x to 2x list at Anthropic and on GPT-5.6 |
| Prompt length | A threshold above which the whole request reprices | OpenAI doubles input above 272,000 tokens; Google and xAI double above 200,000; Anthropic charges flat to 1M |
| Service tier | Batch, flex, standard, fast | Batch and flex are half price; fast and priority run 1.8x to 2.5x |
| Time of day | Peak-hour surcharges | DeepSeek has announced 2x peak pricing but it is not in force |
| Point of sale | First-party API, cloud marketplace, third-party host | [Together AI](https://aiwiki.ai/wiki/together_ai) charges exactly 4x DeepSeek's own rate for the same V4-Pro weights; regional endpoints add 10 percent at OpenAI and Anthropic |
| Reasoning effort | How many output tokens the model spends before answering | 68x on a single prompt within one model family |
| Tokenizer | How many tokens your text becomes | About 30 percent more on Claude 4.7 and later |
| Non-token line items | Server-side tools and session runtime | $10 per 1,000 web searches, $0.08 per session-hour, hourly cache storage at Google |

The rest of this article works through those in order of how much money they tend to move.

## The cross-provider price table

Standard tier, short-context rate, sorted by output price. Cached input is the cache-hit read rate, not the cache-write rate.

| Model | Provider | Input | Cached input | Output |
| --- | --- | --- | --- | --- |
| [DeepSeek V4-Flash](https://aiwiki.ai/wiki/deepseek_v4_flash) | [DeepSeek](https://aiwiki.ai/wiki/deepseek) | $0.14 | $0.0028 | $0.28 |
| GPT-5-nano | [OpenAI](https://aiwiki.ai/wiki/openai) | $0.05 | $0.005 | $0.40 |
| Qwen-Flash (0-256K) | [Alibaba](https://aiwiki.ai/wiki/alibaba) | $0.05 | not published | $0.40 |
| Gemini 2.5 Flash-Lite | [Google](https://aiwiki.ai/wiki/google) | $0.10 | $0.01 | $0.40 |
| [Mistral Small 4](https://aiwiki.ai/wiki/mistral_small_4) | [Mistral AI](https://aiwiki.ai/wiki/mistral_ai) | $0.15 | not published | $0.60 |
| [DeepSeek V4-Pro](https://aiwiki.ai/wiki/deepseek_v4) | DeepSeek | $0.435 | $0.003625 | $0.87 |
| GLM-4.5-Air | [Z.ai](https://aiwiki.ai/wiki/z_ai) | $0.20 | $0.03 | $1.10 |
| GPT-5.6 Luna | OpenAI | $0.20 | $0.02 | $1.20 |
| Qwen-Plus (0-256K) | Alibaba | $0.40 | not published | $1.20 non-thinking, $4.00 thinking |
| GPT-5.4 nano | OpenAI | $0.20 | $0.02 | $1.25 |
| Gemini 3.1 Flash-Lite | Google | $0.25 | $0.025 | $1.50 |
| [Mistral Large 3](https://aiwiki.ai/wiki/mistral_large_3) | Mistral AI | $0.50 | not published | $1.50 |
| GLM-4.7 | Z.ai | $0.60 | $0.11 | $2.20 |
| [Gemini 2.5 Flash](https://aiwiki.ai/wiki/gemini_2_5_flash) | Google | $0.30 | $0.03 | $2.50 |
| Gemini 3.5 Flash-Lite | Google | $0.30 | $0.03 | $2.50 |
| [Grok 4.3](https://aiwiki.ai/wiki/grok_4_3) (under 200K) | [xAI](https://aiwiki.ai/wiki/xai) | $1.25 | $0.20 | $2.50 |
| Gemini 3 Flash Preview | Google | $0.50 | $0.05 | $3.00 |
| GLM-5 | Z.ai | $1.00 | $0.20 | $3.20 |
| [GLM-5.2](https://aiwiki.ai/wiki/glm_5_2) | Z.ai | $1.40 | $0.26 | $4.40 |
| GPT-5.4 mini | OpenAI | $0.75 | $0.075 | $4.50 |
| [Claude Haiku 4.5](https://aiwiki.ai/wiki/claude_haiku_4_5) | [Anthropic](https://aiwiki.ai/wiki/anthropic) | $1.00 | $0.10 | $5.00 |
| [Qwen3-Max](https://aiwiki.ai/wiki/qwen3_max) (0-32K) | Alibaba | $1.20 | not published | $6.00 |
| [Grok 4.5](https://aiwiki.ai/wiki/grok_4_5) (under 200K) | xAI | $2.00 | $0.30 | $6.00 |
| [Mistral Medium 3.5](https://aiwiki.ai/wiki/mistral_medium_3_5) | Mistral AI | $1.50 | not published | $7.50 |
| [Gemini 3.6 Flash](https://aiwiki.ai/wiki/gemini_3_6_flash) | Google | $1.50 | $0.15 | $7.50 |
| Qwen3.7-Max | Alibaba | $2.50 | not published | $7.50 |
| [Gemini 3.5 Flash](https://aiwiki.ai/wiki/gemini_3_5_flash) | Google | $1.50 | $0.15 | $9.00 |
| [Gemini 2.5 Pro](https://aiwiki.ai/wiki/gemini_2_5_pro) (up to 200K) | Google | $1.25 | $0.125 | $10.00 |
| [Claude Sonnet 5](https://aiwiki.ai/wiki/claude_sonnet_5) (to Aug 31, 2026) | Anthropic | $2.00 | $0.20 | $10.00 |
| GPT-5.6 Terra | OpenAI | $2.00 | $0.20 | $12.00 |
| [Gemini 3.1 Pro](https://aiwiki.ai/wiki/gemini_3_1_pro) (up to 200K) | Google | $2.00 | $0.20 | $12.00 |
| GPT-5.4 | OpenAI | $2.50 | $0.25 | $15.00 |
| [Claude Sonnet 4.6](https://aiwiki.ai/wiki/claude_sonnet_4_6) | Anthropic | $3.00 | $0.30 | $15.00 |
| [Kimi K3](https://aiwiki.ai/wiki/kimi_k3) | [Moonshot AI](https://aiwiki.ai/wiki/moonshot_ai) | $3.00 | $0.30 | $15.00 |
| [Claude Opus 5](https://aiwiki.ai/wiki/claude_opus_5) | Anthropic | $5.00 | $0.50 | $25.00 |
| GPT-5.5 | OpenAI | $5.00 | $0.50 | $30.00 |
| GPT-5.6 Sol | OpenAI | $5.00 | $0.50 | $30.00 |
| [Claude Fable 5](https://aiwiki.ai/wiki/claude_fable_5) | Anthropic | $10.00 | $1.00 | $50.00 |
| GPT-5.5 Pro | OpenAI | $30.00 | not published | $180.00 |

Alibaba publishes a third rate for its reasoning models that most comparisons omit. Qwen-Plus bills chain-of-thought output separately from the answer, at $4.00 per 1M tokens in the 0-256K band and $12.00 in the 256K-1M band, so a reasoning-mode request costs materially more than the non-thinking output rate suggests. Qwen3.7-Max also carried a limited-time 50 percent discount when these rates were checked on August 1, 2026.

Sources by provider: OpenAI [1], Anthropic [6], Google [7], DeepSeek [8], xAI [9], Mistral [10], Alibaba [11], Z.ai [12], Moonshot [13].

## Provider-by-provider detail

### OpenAI

OpenAI's current families are GPT-5.6 (Luna, Terra, Sol), the GPT-5.5 line, and the GPT-5.4 line. Cache reads are 90 percent off the base input rate. GPT-5.6 also charges for cache *writes* at 1.25x the uncached input rate, which is new for OpenAI and matches long-standing Anthropic practice.[1][3][4]

| Model | Input | Cached input | Output |
| --- | --- | --- | --- |
| gpt-5.6-sol | $5.00 | $0.50 | $30.00 |
| gpt-5.6-terra | $2.00 | $0.20 | $12.00 |
| gpt-5.6-luna | $0.20 | $0.02 | $1.20 |
| gpt-5.5 | $5.00 | $0.50 | $30.00 |
| gpt-5.5-pro | $30.00 | not published | $180.00 |
| gpt-5.5-cyber | $12.50 | $1.25 | $75.00 |
| gpt-5.4 | $2.50 | $0.25 | $15.00 |
| gpt-5.4-mini | $0.75 | $0.075 | $4.50 |
| gpt-5.4-nano | $0.20 | $0.02 | $1.25 |
| gpt-5.4-pro | $30.00 | not published | $180.00 |
| gpt-5.3-codex | $1.75 | $0.175 | $14.00 |
| gpt-5.2 | $1.75 | $0.175 | $14.00 |
| gpt-5.1 | $1.25 | $0.125 | $10.00 |
| gpt-5 | $1.25 | $0.125 | $10.00 |
| gpt-5-mini | $0.25 | $0.025 | $2.00 |
| gpt-5-nano | $0.05 | $0.005 | $0.40 |
| gpt-5-pro | $15.00 | not published | $120.00 |

Three tier modifiers apply on top. Batch and flex both bill at half the standard rate; fast mode (renamed from priority processing on July 30, 2026) bills at double on the GPT-5.6 and GPT-5.4 lines, and at 2.5 times standard on gpt-5.5. Regional processing endpoints for data residency add a 10 percent uplift on eligible models released on or after March 5, 2026.[1][5]

| Model | Standard in/out | Batch and flex in/out | Fast mode in/out |
| --- | --- | --- | --- |
| gpt-5.6-sol | $5.00 / $30.00 | $2.50 / $15.00 | $10.00 / $60.00 |
| gpt-5.6-terra | $2.00 / $12.00 | $1.00 / $6.00 | $4.00 / $24.00 |
| gpt-5.6-luna | $0.20 / $1.20 | $0.10 / $0.60 | $0.40 / $2.40 |
| gpt-5.5 | $5.00 / $30.00 | $2.50 / $15.00 | $12.50 / $75.00 |
| gpt-5.4 | $2.50 / $15.00 | $1.25 / $7.50 | $5.00 / $30.00 |

### Anthropic

Anthropic publishes four prices per model rather than three, because a cache write is billed separately from a cache read and the write price depends on the cache lifetime. A 5-minute write costs 1.25x the base input rate, a 1-hour write costs 2x, and a read costs 0.1x.[6]

| Model | Input | 5m cache write | 1h cache write | Cache hit | Output |
| --- | --- | --- | --- | --- | --- |
| [Claude Fable 5](https://aiwiki.ai/wiki/claude_fable_5) | $10 | $12.50 | $20 | $1.00 | $50 |
| [Claude Mythos 5](https://aiwiki.ai/wiki/claude_mythos_5) (limited availability) | $10 | $12.50 | $20 | $1.00 | $50 |
| Claude Opus 5 | $5 | $6.25 | $10 | $0.50 | $25 |
| [Claude Opus 4.8](https://aiwiki.ai/wiki/claude_opus_4_8) | $5 | $6.25 | $10 | $0.50 | $25 |
| Claude Opus 4.7 | $5 | $6.25 | $10 | $0.50 | $25 |
| Claude Opus 4.6 | $5 | $6.25 | $10 | $0.50 | $25 |
| Claude Opus 4.5 | $5 | $6.25 | $10 | $0.50 | $25 |
| Claude Sonnet 5 (through Aug 31, 2026) | $2 | $2.50 | $4 | $0.20 | $10 |
| Claude Sonnet 5 (from Sep 1, 2026) | $3 | $3.75 | $6 | $0.30 | $15 |
| Claude Sonnet 4.6 | $3 | $3.75 | $6 | $0.30 | $15 |
| Claude Sonnet 4.5 | $3 | $3.75 | $6 | $0.30 | $15 |
| Claude Haiku 4.5 | $1 | $1.25 | $2 | $0.10 | $5 |

Anthropic is the one major provider with no long-context surcharge: Claude 4.6 and later include the full 1,000,000-token window at standard rates, and the docs state that "a 900k-token request is billed at the same per-token rate as a 9k-token request."[6] The Batch API halves both input and output. A fast mode research preview prices Claude Opus 5 and Opus 4.8 at $10 and $50, double standard, and is not combinable with batch. Requesting US-only inference with the `inference_geo` parameter on Claude 4.6 and later multiplies every token category by 1.1.[6]

Two line items sit outside the token price. Server-side web search is $10 per 1,000 searches on top of token costs. Claude Managed Agents adds $0.08 per session-hour of running time, metered only while a session is actually running.[6]

### Google

Google publishes four service tiers per model. Batch and flex are half the standard rate; priority is 1.8x. Context caching carries both a per-token read rate and an hourly storage charge, which is unusual and easy to miss when a cache sits idle.[7]

| Model | Input | Output | Cache read | Cache storage |
| --- | --- | --- | --- | --- |
| Gemini 3.6 Flash | $1.50 | $7.50 | $0.15 | $1.00/hr |
| Gemini 3.5 Flash | $1.50 | $9.00 | $0.15 | $1.00/hr |
| Gemini 3.5 Flash-Lite | $0.30 | $2.50 | $0.03 | $1.00/hr |
| Gemini 3.1 Pro Preview (up to 200K) | $2.00 | $12.00 | $0.20 | $4.50/hr |
| Gemini 3.1 Pro Preview (over 200K) | $4.00 | $18.00 | $0.40 | $4.50/hr |
| Gemini 3.1 Flash-Lite | $0.25 | $1.50 | $0.025 | $1.00/hr |
| Gemini 3 Flash Preview | $0.50 | $3.00 | $0.05 | $1.00/hr |
| Gemini 2.5 Pro (up to 200K) | $1.25 | $10.00 | $0.125 | $4.50/hr |
| Gemini 2.5 Pro (over 200K) | $2.50 | $15.00 | $0.25 | $4.50/hr |
| Gemini 2.5 Flash | $0.30 | $2.50 | $0.03 | $1.00/hr |
| Gemini 2.5 Flash-Lite | $0.10 | $0.40 | $0.01 | $1.00/hr |

Audio input is priced above text, image, and video input on the Flash and Flash-Lite models: $1.00 against $0.30 on Gemini 2.5 Flash, $0.50 against $0.25 on Gemini 3.1 Flash-Lite, and $0.30 against $0.10 on Gemini 2.5 Flash-Lite.[7] As of August 1, 2026 the pricing page lists no Pro-tier model above Gemini 3.1 Pro Preview.

### DeepSeek

DeepSeek serves two models and publishes three prices for each.[8]

| Model | Cache-hit input | Cache-miss input | Output |
| --- | --- | --- | --- |
| `deepseek-v4-flash` | $0.0028 | $0.14 | $0.28 |
| `deepseek-v4-pro` | $0.003625 | $0.435 | $0.87 |

The cache-hit rate is the outlier of the whole market. At $0.0028 against $0.14, a hit costs 2 percent of a miss, where the industry norm is 10 percent. [Artificial Analysis](https://aiwiki.ai/wiki/artificial_analysis) called this out directly on July 31, describing a "~98% cache hit discount on its first-party API" against the 90 percent standard elsewhere.[25]

### xAI

xAI prices every model in two bands split at 200,000 tokens, with the upper band exactly double the lower on all three rates.[9]

| Model | Context | Input (under 200K) | Cached (under 200K) | Output (under 200K) | Input (200K+) | Output (200K+) |
| --- | --- | --- | --- | --- | --- | --- |
| Grok 4.5 | 500K | $2.00 | $0.30 | $6.00 | $4.00 | $12.00 |
| Grok 4.3 | 1M | $1.25 | $0.20 | $2.50 | $2.50 | $5.00 |
| Grok 4.20-0309 (reasoning) | 1M | $1.25 | $0.20 | $2.50 | $2.50 | $5.00 |
| Grok 4.20-0309 (non-reasoning) | 1M | $1.25 | $0.20 | $2.50 | $2.50 | $5.00 |
| Grok 4.20 multi-agent 0309 | 1M | $1.25 | $0.20 | $2.50 | $2.50 | $5.00 |
| Grok build 0.1 | 256K | $1.00 | $0.20 | $2.00 | $2.00 | $4.00 |

### Mistral AI

Mistral does not publish per-model cache-hit rates, stating a flat 90 percent saving on cached input tokens, and offers 50 percent off for batch processing.[10]

| Model | Input | Output |
| --- | --- | --- |
| Mistral Medium 3.5 | $1.50 | $7.50 |
| Mistral Large 3 | $0.50 | $1.50 |
| Mistral Small 4 | $0.15 | $0.60 |
| Ministral 3 (3B, 8B, 14B) | $0.10 to $0.20 | $0.10 to $0.20 |
| Magistral Medium | $2.00 | $5.00 |
| Magistral Small | $0.50 | $1.50 |
| Codestral | $0.30 | $0.90 |
| Devstral 2 | $0.40 | $2.00 |

The Ministral 3 line is priced symmetrically, the same rate for input and output, which is rare and makes it unusually attractive for output-heavy chat where responses run long.

### Alibaba (Qwen)

Alibaba's international Model Studio endpoint prices most Qwen models in input-length bands, which is a different mechanism from OpenAI's or Google's single threshold: the band is chosen by the request's input length and applies to that request.[11]

| Model | Input band | Input | Output |
| --- | --- | --- | --- |
| Qwen3.7-Max | 0 to 1M | $2.50 | $7.50 |
| Qwen3-Max | 0 to 32K | $1.20 | $6.00 |
| Qwen3-Max | 32K to 128K | $2.40 | $12.00 |
| Qwen3-Max | 128K to 256K | $3.00 | $15.00 |
| Qwen-Plus | 0 to 256K | $0.40 | $1.20 |
| Qwen-Plus | 256K to 1M | $1.20 | $3.60 |
| Qwen-Flash | 0 to 256K | $0.05 | $0.40 |
| Qwen-Flash | 256K to 1M | $0.25 | $2.00 |

Qwen-Flash at $0.05 input ties GPT-5-nano for the lowest published input rate of any model in this article. Note that earlier revisions of this page listed Qwen3-Max at $0.86 and $3.44; those figures do not match Alibaba's current international price list and have been corrected.

### Z.ai (GLM)

| Model | Input | Cached input | Output |
| --- | --- | --- | --- |
| GLM-5.2 | $1.40 | $0.26 | $4.40 |
| GLM-5 | $1.00 | $0.20 | $3.20 |
| GLM-4.7 | $0.60 | $0.11 | $2.20 |
| GLM-4.5-Air | $0.20 | $0.03 | $1.10 |

Z.ai lists cached-input storage fees as free for a limited time on all four models.[12]

### Moonshot AI (Kimi)

Moonshot prices Kimi K3 flat across its full 1,048,576-token window, with the discounted input rate applied automatically when prefix caching hits.[13]

| Model | Cache-hit input | Uncached input | Output |
| --- | --- | --- | --- |
| Kimi K3 | $0.30 | $3.00 | $15.00 |

## The July 30 price cut: Luna down 80 percent, Terra down 20 percent

The most consequential pricing event of the past week was OpenAI's July 30 repricing. The API changelog records it plainly: "Starting July 30, GPT-5.6 Luna costs 80% less, while GPT-5.6 Terra costs 20% less."[2] The pricing page now lists Luna at $0.20 input, $0.02 cached input, and $1.20 output, against the $1.00 / $0.10 / $6.00 it carried at launch three weeks earlier. Terra fell from $2.50 / $0.25 / $15.00 to $2.00 / $0.20 / $12.00. GPT-5.6 Sol was not cut and remains $5.00 / $0.50 / $30.00.[1][2][17]

The shape of the cut is as interesting as its size. OpenAI repriced the two cheaper tiers and left the flagship untouched, which widened the spread inside a single model family from 5x to 25x on both input and output. Nothing else changed: the long-context surcharge above 272,000 input tokens is still 2x input and 1.5x output, cache writes are still 1.25x the uncached input rate, and rate limits were not adjusted.[1][3]

The effect on the rankings is real. Luna is now cheaper on output than GPT-5.4 nano, a model two tiers below it in capability, and cheaper on both rates than [Gemini 2.5 Flash](https://aiwiki.ai/wiki/gemini_2_5_flash). At one twenty-fifth of Sol's headline price, a model that Artificial Analysis scored at 51 on its Intelligence Index now sits in the same price band as the industry's smallest models.[18][20]

A second pricing change shipped the same day, covered under service tiers below.

## Why headline per-token prices mislead

This is the part naive comparisons get wrong, and it now dominates the arithmetic.

Reasoning models spend output tokens thinking before they answer, and how many they spend is a setting, not a constant. Willison, who had early access to GPT-5.6, priced one identical SVG-drawing prompt across all three models at six effort levels. The cheapest run was Luna at effort `none` for **0.71 cents**. The most expensive was Sol at `max` for **48.55 cents**. Same prompt, same family, same week, a spread of about 68 times, most of it from token counts rather than rates. His conclusion: "price-per-million tokens doesn't tell us much now that the number of reasoning tokens can differ so much between models for the same task."[17] Those figures predate the July 30 cut, so the Luna end of that range is now lower still.

The better comparison is cost per task. Artificial Analysis measures the full dollar cost of running its Intelligence Index on each model, computing each evaluation's cost "from input, cache hit, cache write, reasoning, and answer token prices, divided by task count, and weighted by its Intelligence Index weight", and using the token counts each provider's own API reports rather than a standardized client-side tokenizer.[24][27] The figures below are from Artificial Analysis model pages as listed on August 1, 2026, all at maximum reasoning effort.

| Model | Intelligence Index | Cost to run the full index | Blended rate (7:2:1) |
| --- | --- | --- | --- |
| Claude Fable 5 | 60 | $5,630.52 | $7.70 |
| GPT-5.6 Sol | 59 | $3,442.81 | $4.35 |
| GPT-5.6 Luna | 51 | $190.87 | $0.17 |
| DeepSeek V4-Flash 0731 | 50 | $72.02 | $0.06 |
| DeepSeek V4-Pro | 44 | $176.34 | $0.18 |

Per-model sources: Claude Fable 5 [23], GPT-5.6 Sol [19], GPT-5.6 Luna [20], DeepSeek V4-Flash [21], DeepSeek V4-Pro [22]. The blended rate is Artificial Analysis's own 7:2:1 weighting of cache-hit, input, and output tokens.

Three things fall out of that table that a per-token comparison hides.

**Cost per task compresses the headline gap.** Sol's list rates are 25 times Luna's on both input and output, but the measured cost of running the same benchmark suite differed by about 18 times. Luna talks more: Artificial Analysis noted it generated 130 million output tokens over the evaluation.[19][20] Verbosity ate roughly a quarter of the apparent saving.

**Cost per task can invert the headline ranking.** DeepSeek V4-Pro carries a higher list price than V4-Flash and scored six points lower on the index, yet cost more than twice as much to evaluate, because its per-token rates are roughly three times higher. It was not the more verbose of the two: Artificial Analysis recorded 180 million generated tokens for V4-Pro against 210 million for V4-Flash-0731.[21][22]

**Similar scores can sit at very different prices.** Artificial Analysis put V4-Flash 0731's cost per task at "~60% lower than GPT-5.6 Luna (max), a model with comparable intelligence", and that was measured the day after OpenAI's 80 percent cut.[25] The full-index figures agree: $72.02 against $190.87, a 62 percent difference, for scores of 50 and 51. Within days that spread was general business-press news: Bloomberg reported on August 4, 2026, citing Artificial Analysis benchmark testing, that executing a complex real-world workload cost $0.03 with DeepSeek V4-Flash against $3.15 with Claude Fable 5, and wrote that a so-called DeepSeek death zone had emerged on a widely shared Artificial Analysis chart for rivals that charge more for the same capability or deliver less for the same price.[31]

For a sense of the top of the range, Artificial Analysis reported in June that [Claude Fable 5](https://aiwiki.ai/wiki/claude_fable_5) was "the most expensive model we have ever benchmarked", at roughly 1.7 times the next-highest model, Claude Opus 4.8, and 2.2 times GPT-5.5 at extra-high effort.[26] The July article on the GPT-5.6 launch gave per-task costs of $1.04 for Sol at max effort, $0.55 for Terra, and $0.21 for Luna, and observed that Sol offered "a similar level of intelligence to Claude Fable 5 at approximately one third of the cost."[18][29] Those per-task figures also predate the Luna cut.

The practical upshot: benchmark your own prompt at the effort level you will actually ship, and compare total spend. A rate card cannot tell you how talkative a model is.

## Cache pricing is now a first-order cost

[Context caching](https://aiwiki.ai/wiki/context_caching) has moved from a nice-to-have to one of the largest levers on a real bill, especially for agents that resend a growing conversation on every turn. Three separate prices are involved, and providers differ on all three.

| Provider | Cache-hit discount | Cache-write charge | Storage charge |
| --- | --- | --- | --- |
| DeepSeek | ~98 percent ($0.0028 vs $0.14 on V4-Flash) | none published | none published |
| OpenAI | 90 percent | 1.25x uncached input on GPT-5.6 | none published |
| Anthropic | 90 percent (0.1x base input) | 1.25x for 5 minutes, 2x for 1 hour | none published |
| Google | 90 percent | none published | $1.00 to $4.50 per hour depending on model |
| xAI | 85 percent on Grok 4.5, 84 percent on Grok 4.3 | none published | none published |
| Mistral | 90 percent (stated, not per model) | none published | none published |
| Z.ai | 80 to 85 percent depending on model | none published | free for a limited time |
| Moonshot | 90 percent | none published | none published |

Sources: [6][7][8][9][10][12][13].

Two consequences. First, DeepSeek's cache economics are structurally different from everyone else's, not marginally better: at 98 percent off, resending a 100,000-token prefix costs $0.28 instead of $14.00. Artificial Analysis attributed much of DeepSeek's cost-per-task advantage to exactly this.[25] Second, a cache is not free to fill. Anthropic's own guidance is that a 5-minute cache pays for itself after one read and a 1-hour cache after two, which means a cache with a low hit rate can cost more than no cache at all.[6] Google's hourly storage charge creates the same trap in a different shape: an idle cached context on Gemini 3.1 Pro accrues $4.50 an hour whether or not anything reads it.[7] See [prompt caching](https://aiwiki.ai/wiki/prompt_caching) for the mechanics.

## Long-context surcharges

Several providers reprice an entire request once its prompt crosses a threshold. This is not a marginal rate on the tokens above the line; at OpenAI it applies to the full request.

| Provider | Threshold | What changes |
| --- | --- | --- |
| OpenAI (GPT-5.6, 5.5, 5.4) | 272,000 input tokens | 2x input and 1.5x output for the full request[3][4] |
| Google (Gemini 3.1 Pro, 2.5 Pro) | 200,000 tokens | Input and cache roughly double, output rises 1.5x on 3.1 Pro and 1.5x on 2.5 Pro[7] |
| xAI (all models) | 200,000 tokens | All three rates double[9] |
| Alibaba (Qwen) | 32K, 128K, 256K bands by model | Rates step up per band[11] |
| Anthropic (Claude 4.6+) | none | Full 1M window at standard rates[6] |
| DeepSeek, Moonshot | none published | Flat to 1M tokens[8][13] |

The practical reading: Anthropic and DeepSeek are the cheapest places to put a genuinely enormous prompt relative to their own short-context rates, while a GPT-5.6 Sol request at 300,000 input tokens is billed at $10 input and $45 output, not $5 and $30. For window sizes rather than prices, see [LLM context window comparison](https://aiwiki.ai/wiki/llm_context_window_comparison).

## Batch, flex, and premium tiers

Nearly every provider sells the same model at three or four latency classes.

| Tier | Typical price | Trade-off |
| --- | --- | --- |
| Batch | 50 percent of standard at OpenAI, Anthropic, Google, and Mistral | Asynchronous, results in hours[1][6][7][10] |
| Flex | 50 percent of standard at OpenAI and Google | Synchronous but slower, with occasional 429 resource-unavailable responses[5][7] |
| Standard | list | baseline |
| Fast or priority | 2x at OpenAI and Anthropic, 1.8x at Google | Lower latency[1][6][7] |

OpenAI now has the cleanest version of this ladder in the market, and the second half of its July 30 changes built it. Priority processing was renamed Fast mode that day, a rename recorded in a footnote on the pricing page and repeated in the guide: "Priority processing was renamed Fast mode on July 30, 2026."[1][30] Fast mode is exactly double standard on every GPT-5.6 tier, giving a four-step ladder at 0.5x, 0.5x, 1x, and 2x of list:

| Model | Batch | Flex | Standard | Fast mode |
| --- | --- | --- | --- | --- |
| gpt-5.6-sol | $2.50 / $15.00 | $2.50 / $15.00 | $5.00 / $30.00 | $10.00 / $60.00 |
| gpt-5.6-terra | $1.00 / $6.00 | $1.00 / $6.00 | $2.00 / $12.00 | $4.00 / $24.00 |
| gpt-5.6-luna | $0.10 / $0.60 | $0.10 / $0.60 | $0.20 / $1.20 | $0.40 / $2.40 |

The documented speed gain is "up to 2.5x faster than Standard processing" for `gpt-5.6-sol`; OpenAI publishes no figure for Terra or Luna. The rename is backward compatible, and either `service_tier: "priority"` or `service_tier: "fast"` selects the tier.[30]

OpenAI's flex tier bills "at Batch API rates, with additional discounts from prompt caching", is in beta with limited model availability, and needs a raised client timeout because request timeouts are more likely; requests that fail with 429 are not charged.[5] Anthropic's fast mode is a research preview limited to Claude Opus 5 and Opus 4.8, is first-party only, and cannot be combined with the Batch API.[6] For any non-interactive job, evaluations, bulk extraction, [data labeling](https://aiwiki.ai/wiki/data_labeling), the batch tier halves every number in this article. See the [OpenAI Batch API](https://aiwiki.ai/wiki/batch_api) page for mechanics.

## Peak and off-peak pricing

DeepSeek's pricing page carries a note that the API "will soon adopt a peak/off-peak pricing policy", under which "during peak hours, prices will be 2x the regular prices, applicable to all billing items", with peak hours defined as 9:00 to 12:00 and 14:00 to 18:00 Beijing time (UTC+8) daily.[8]

**This is not in force.** The same note states that "the effective date will be subject to the official announcement", and as of August 5, 2026 no start date has been published. The rates in the DeepSeek table above are the ones actually charged at every hour of the day. Any comparison that applies a 2x peak multiplier to DeepSeek today is wrong.

No other provider covered here publishes time-of-day pricing.

## Open weights change what an API price means

For a model published under a permissive license, the API rate is a price for a service, not a price for the model. [DeepSeek V4](https://aiwiki.ai/wiki/deepseek_v4) is the clearest case: DeepSeek-V4-Pro carries an MIT license tag on Hugging Face, with 1.6 trillion total parameters, 49 billion active, and a one-million-token [context window](https://aiwiki.ai/wiki/context_window).[28] You can run it on your own hardware and pay nothing per token, or buy it from someone other than DeepSeek.

The spread between hosts for identical weights is large:

| Model | First-party API | Third-party host |
| --- | --- | --- |
| DeepSeek V4-Pro | $0.435 / $0.87 (DeepSeek)[8] | $1.74 / $3.48, cached $0.20 ([Together AI](https://aiwiki.ai/wiki/together_ai))[14] |
| Kimi K3 | $3.00 / $15.00 (Moonshot)[13] | $3.00 / $15.00, cached $0.30 (Together AI)[14] |
| GLM-5.2 | $1.40 / $4.40 (Z.ai)[12] | $1.40 / $4.40, cached $0.26 (Together AI)[14] |

Together AI charges exactly four times DeepSeek's own rate for V4-Pro, while matching first-party pricing on Kimi K3 and GLM-5.2. Speed-oriented hosts price differently again. [Groq](https://aiwiki.ai/wiki/groq) lists GPT-OSS 120B at $0.15 and $0.60, GPT-OSS 20B at $0.075 and $0.30, Llama 3.1 8B Instant at $0.05 and $0.08, Llama 3.3 70B Versatile at $0.59 and $0.79, and Qwen 3.6 27B at $0.60 and $3.00.[15] [Amazon Bedrock](https://aiwiki.ai/wiki/amazon_bedrock) prices its own [Amazon Nova](https://aiwiki.ai/wiki/amazon_nova) family at $0.035 and $0.14 for Nova Micro, $0.06 and $0.24 for Nova Lite, and $0.80 and $3.20 for Nova Pro in the US East and US West regions, and hosts third-party open models such as DeepSeek V3.2 at $0.62 and $1.85 and Qwen3 Next 80B at $0.15 and $1.20.[16] Nova Micro's $0.035 input rate is the lowest per-token input price of any model named in this article.

Aggregators including [OpenRouter](https://aiwiki.ai/wiki/openrouter) resell many of these endpoints, and their listed prices track the underlying host rather than the model. Availability, rate limits, and feature parity move independently of the model release, so a price quoted for "DeepSeek V4-Pro" is meaningless without naming who is serving it.

## Tokenizers, and other reasons per-token rates are not comparable

A per-token price is only comparable if the token counts are. Anthropic states that Claude 4.7 and later models, plus Claude Mythos Preview, use a newer tokenizer that "produces approximately 30% more tokens for the same text", with the exact increase depending on content and workload; Claude Sonnet 4.6 and earlier use the previous tokenizer.[6] That erodes a meaningful part of any apparent saving from a lower headline rate on a newer Claude model, and it means comparing Opus 5 against Sonnet 4.6 on rate alone understates Opus 5's cost. Artificial Analysis handles the same problem by using each provider's reported token counts for cost reporting while standardizing on a single tokenizer for its speed measurements.[27] See [tokenization](https://aiwiki.ai/wiki/tokenization) for the underlying mechanics.

Two smaller distortions are worth naming. Output tokens run about 2x to 8x input across the models in this article, so output price dominates chat and generation workloads while input price dominates retrieval, classification, and long-document reading; that is why the main table is sorted by output. And several charges sit entirely outside the token meter, including Anthropic's $10 per 1,000 web searches and $0.08 per session-hour for Managed Agents, and Google's hourly cache storage.[6][7]

## How these prices were verified

Every figure was read on August 1, 2026 from the provider's own pricing documentation: OpenAI [1], Anthropic [6], Google [7], DeepSeek [8], xAI [9], Mistral [10], Alibaba Cloud Model Studio [11], Z.ai [12], Moonshot [13], Together AI [14], Groq [15], and Amazon Web Services [16]. Cost-per-task and cost-to-run figures come from [Artificial Analysis](https://aiwiki.ai/wiki/artificial_analysis) model pages and articles, dated where the date matters.

No number here was carried forward from the previous revision without being re-checked, and the check found real drift: Qwen3-Max was listed at $0.86 and $3.44 against Alibaba's current $1.20 and $6.00 base band; Grok 4.3 was shown without its published $0.20 cache rate and without the 200,000-token doubling; Grok 4.1 Fast has since disappeared from xAI's model page; and the whole GPT-5.6 family, Claude Opus 5, Claude Sonnet 5, Gemini 3.6 Flash, Kimi K3, and the GLM line were absent. Third-party aggregators were used only to locate official pages, never as a price source. Where a current price could not be confirmed on an official page, the model was omitted.

Related pages: [AI pricing](https://aiwiki.ai/wiki/ai_pricing) for subscription and market economics, [inference](https://aiwiki.ai/wiki/inference) for what you are actually paying for, [frontier models](https://aiwiki.ai/wiki/frontier_models) for the capability tiers referenced here, and [LLM benchmark comparison](https://aiwiki.ai/wiki/llm_benchmark_comparison) for the capability side of the price-performance question.

## References

1. OpenAI, "Pricing," OpenAI API docs (accessed August 1, 2026). https://developers.openai.com/api/docs/pricing
2. OpenAI, "Changelog," OpenAI API docs (accessed August 1, 2026). https://developers.openai.com/api/docs/changelog
3. OpenAI, "gpt-5.6-luna," OpenAI API model reference (accessed August 1, 2026). https://developers.openai.com/api/docs/models/gpt-5.6-luna
4. OpenAI, "gpt-5.6," OpenAI API model reference (accessed August 1, 2026). https://developers.openai.com/api/docs/models/gpt-5.6
5. OpenAI, "Flex processing," OpenAI API docs (accessed August 1, 2026). https://developers.openai.com/api/docs/guides/flex-processing
6. Anthropic, "Pricing," Claude Platform docs (accessed August 1, 2026). https://platform.claude.com/docs/en/about-claude/pricing
7. Google, "Gemini Developer API pricing," Google AI for Developers (accessed August 1, 2026). https://ai.google.dev/gemini-api/docs/pricing
8. DeepSeek, "Models & Pricing," DeepSeek API docs (accessed August 1, 2026). https://api-docs.deepseek.com/quick_start/pricing
9. xAI, "Models," xAI docs (accessed August 1, 2026). https://docs.x.ai/developers/models
10. Mistral AI, "API pricing," La Plateforme (accessed August 1, 2026). https://mistral.ai/pricing/api
11. Alibaba Cloud, "Model pricing," Model Studio documentation (accessed August 1, 2026). https://www.alibabacloud.com/help/en/model-studio/model-pricing
12. Z.ai, "Pricing," Z.ai developer guides (accessed August 1, 2026). https://docs.z.ai/guides/overview/pricing
13. Kimi, "Kimi K3 Pricing Explained: Plans and API Costs" (accessed August 1, 2026). https://www.kimi.com/resources/kimi-k3-pricing
14. Together AI, "Pricing" (accessed August 1, 2026). https://www.together.ai/pricing
15. Groq, "Pricing" (accessed August 1, 2026). https://groq.com/pricing
16. Amazon Web Services, "Amazon Bedrock pricing" (accessed August 1, 2026). https://aws.amazon.com/bedrock/pricing/
17. Simon Willison, "The new GPT-5.6 family: Luna, Terra, Sol," Simon Willison's Weblog, July 9, 2026. https://simonwillison.net/2026/Jul/9/gpt-5-6/
18. Artificial Analysis, "GPT-5.6 has landed," July 9, 2026. https://artificialanalysis.ai/articles/gpt-5-6-has-landed
19. Artificial Analysis, "GPT-5.6 Sol: Intelligence, Performance & Price Analysis" (accessed August 1, 2026). https://artificialanalysis.ai/models/gpt-5-6-sol
20. Artificial Analysis, "GPT-5.6 Luna: Intelligence, Performance & Price Analysis" (accessed August 1, 2026). https://artificialanalysis.ai/models/gpt-5-6-luna
21. Artificial Analysis, "DeepSeek V4 Flash: Intelligence, Performance & Price Analysis" (accessed August 1, 2026). https://artificialanalysis.ai/models/deepseek-v4-flash
22. Artificial Analysis, "DeepSeek V4 Pro: Intelligence, Performance & Price Analysis" (accessed August 1, 2026). https://artificialanalysis.ai/models/deepseek-v4-pro
23. Artificial Analysis, "Claude Fable 5: Intelligence, Performance & Price Analysis" (accessed August 1, 2026). https://artificialanalysis.ai/models/claude-fable-5
24. Artificial Analysis, "Models" index (accessed August 1, 2026). https://artificialanalysis.ai/models
25. Artificial Analysis, "DeepSeek V4 Flash 0731 scores 50 on the Artificial Analysis Intelligence Index," @ArtificialAnlys on X, July 31, 2026. https://x.com/ArtificialAnlys/status/2083123180869496865
26. Artificial Analysis, "Claude Fable 5 cost ~$6.2K to run the Artificial Analysis Intelligence Index benchmarks," @ArtificialAnlys on X, June 17, 2026. https://x.com/ArtificialAnlys/status/2067384319942029379
27. Artificial Analysis, "Intelligence benchmarking methodology" (accessed August 1, 2026). https://artificialanalysis.ai/methodology/intelligence-benchmarking
28. DeepSeek, "DeepSeek-V4-Pro," Hugging Face model repository (accessed August 1, 2026). https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro
29. Artificial Analysis, "How GPT-5.6 Sol, Terra, Luna compare on intelligence vs cost," July 13, 2026. https://artificialanalysis.ai/articles/gpt-5-6-intelligence-vs-cost-across-sol-terra-luna
30. OpenAI, "Fast mode," OpenAI API docs (accessed August 1, 2026). https://developers.openai.com/api/docs/guides/fast-mode
31. Zheping Huang, Nectar Gan and Saritha Rai, "China's AI Blitz Creates 'Death Zone' for Rival US Model Makers," Bloomberg News, August 4, 2026. https://www.bloomberg.com/news/articles/2026-08-04/china-s-ai-blitz-creates-death-zone-for-rival-us-model-makers