by Inference Hub

How AI API Prices Changed, 2023–2026: Token Prices in Charts

Official per-token prices for every generation of GPT, Claude and Gemini, charted: where flagship prices fell, where small models got more expensive, and the price rises already scheduled.

pricingllmopenaianthropicgoogledata

AI API prices haven’t fallen in a straight line. Flagship models are much cheaper than in 2023: GPT-4 cost $30 input / $60 output per million tokens, and GPT-6 Sol costs $2 / $10. But prices rose for long stretches in between, and both OpenAI and Anthropic now sell a $10 / $50 tier above their old flagship. Small models went the other way: Gemini Flash costs ten times what Gemini 1.5 Flash did. Google is raising prices again this winter, while OpenAI’s and Anthropic’s newest small models undercut every earlier mini and Haiku model.

All prices below are official list prices per 1M tokens from each lab’s own API, checked on November 17, 2026. They are the standard real-time rate for the shortest context band, without Batch or caching discounts. Live prices for every model are on our model pages, and every price change we record is on the changelog.

Flagship models: cheaper than 2023, but not every year

OpenAI: flagship model list prices by generationGPT-4 $30 input, $60 output; GPT-4 Turbo $10 input, $30 output; GPT-4o $2.50 input, $10 output; GPT-4.1 $2 input, $8 output; GPT-5 $1.25 input, $10 output; GPT-5.2 $1.75 input, $14 output; GPT-5.4 $2.50 input, $15 output; GPT-5.5 $5 input, $30 output; GPT-5.6 Sol $4 input, $20 output; GPT-6 Sol $2 input, $10 output per 1M tokens.OpenAIInputOutput$1$3$10$30$100GPT-4: $30 input / $60 output per 1M tokensGPT-4$30 / $60GPT-4 Turbo: $10 input / $30 output per 1M tokensGPT-4 TurboGPT-4o: $2.50 input / $10 output per 1M tokensGPT-4oGPT-4.1: $2 input / $8 output per 1M tokensGPT-4.1GPT-5: $1.25 input / $10 output per 1M tokensGPT-5GPT-5.2: $1.75 input / $14 output per 1M tokensGPT-5.2GPT-5.4: $2.50 input / $15 output per 1M tokensGPT-5.4GPT-5.5: $5 input / $30 output per 1M tokensGPT-5.5GPT-5.6 Sol: $4 input / $20 output per 1M tokensGPT-5.6 SolGPT-6 Sol: $2 input / $10 output per 1M tokensGPT-6 Sol$2 / $10
Claude Opus: flagship model list prices by generationClaude 3 Opus $15 input, $75 output; Claude Opus 4 $15 input, $75 output; Claude Opus 4.1 $15 input, $75 output; Claude Opus 4.5 $5 input, $25 output; Claude Opus 4.6 $5 input, $25 output; Claude Opus 4.7 $5 input, $25 output; Claude Opus 4.8 $5 input, $25 output; Claude Opus 5 $5 input, $25 output; Claude Opus 5.5 $4 input, $20 output per 1M tokens.Claude OpusInputOutput$1$3$10$30$100Claude 3 Opus: $15 input / $75 output per 1M tokens3 Opus$15 / $75Claude Opus 4: $15 input / $75 output per 1M tokensOpus 4Claude Opus 4.1: $15 input / $75 output per 1M tokensOpus 4.1Claude Opus 4.5: $5 input / $25 output per 1M tokensOpus 4.5Claude Opus 4.6: $5 input / $25 output per 1M tokensOpus 4.6Claude Opus 4.7: $5 input / $25 output per 1M tokensOpus 4.7Claude Opus 4.8: $5 input / $25 output per 1M tokensOpus 4.8Claude Opus 5: $5 input / $25 output per 1M tokensOpus 5Claude Opus 5.5: $4 input / $20 output per 1M tokensOpus 5.5$4 / $20
Claude Sonnet: flagship model list prices by generationClaude 3.5 Sonnet $3 input, $15 output; Claude Sonnet 4 $3 input, $15 output; Claude Sonnet 4.5 $3 input, $15 output; Claude Sonnet 4.6 $3 input, $15 output; Claude Sonnet 5 $2 input, $10 output; Claude Sonnet 5.5 $2 input, $10 output per 1M tokens.Claude SonnetInputOutput$1$3$10$30$100Claude 3.5 Sonnet: $3 input / $15 output per 1M tokens3.5 Sonnet$3 / $15Claude Sonnet 4: $3 input / $15 output per 1M tokensSonnet 4Claude Sonnet 4.5: $3 input / $15 output per 1M tokensSonnet 4.5Claude Sonnet 4.6: $3 input / $15 output per 1M tokensSonnet 4.6Claude Sonnet 5: $2 input / $10 output per 1M tokensSonnet 5Claude Sonnet 5.5: $2 input / $10 output per 1M tokensSonnet 5.5$2 / $10
Gemini Pro / Gemini 4: flagship model list prices by generationGemini 1.5 Pro $1.25 input, $5 output; Gemini 2.5 Pro $1.25 input, $10 output; Gemini 3 Pro $2 input, $12 output; Gemini 3.1 Pro $2 input, $12 output; Gemini 4 Carbon $2 input, $10 output; Gemini 4 Carbon, Dec 1 $4 input, $20 output per 1M tokens.Gemini Pro / Gemini 4InputOutput$1$3$10$30$100Gemini 1.5 Pro: $1.25 input / $5 output per 1M tokens1.5 Pro$1.25 / $5Gemini 2.5 Pro: $1.25 input / $10 output per 1M tokens2.5 ProGemini 3 Pro: $2 input / $12 output per 1M tokens3 ProGemini 3.1 Pro: $2 input / $12 output per 1M tokens3.1 ProGemini 4 Carbon: $2 input / $10 output per 1M tokens4 CarbonGemini 4 Carbon, Dec 1: $4 input / $20 output per 1M tokens4 Carbon, Dec 1$4 / $20
Official list price per 1M tokens at each generation, standard real-time tier, shortest context band. Log scale. Claude 3 Opus, Claude 3.5 Sonnet, GPT-4, GPT-4 Turbo and Gemini 1.5 Pro are launch-era list prices; the rest are current official prices from our listings.
  • OpenAI cut GPT-4’s $30 / $60 to $2 / $8 by GPT-4.1, then raised prices from generation to generation. GPT-5.5 cost $5 / $30, four times GPT-5’s input price. With GPT-5.6 OpenAI split its lineup into tiers, and GPT-6 Sol is back at $2 / $10. The new top tier, GPT-6 Astra, costs $10 / $50.
  • Anthropic kept Opus at $15 / $75 for three generations, then cut it by two-thirds to $5 / $25 with Opus 4.5. That price held through Opus 5, and Opus 5.5 is $4 / $20. Like OpenAI, Anthropic added a tier on top: Claude Fable, launched at $10 / $50, with Fable 5.2 now $8 / $40.
  • Sonnet stayed at $3 / $15 from Claude 3.5 Sonnet to Sonnet 4.6. Sonnet 5 launched at $2 / $10, and Anthropic cancelled the rise back to $3 / $15 it had planned for September 1, 2026.
  • Google’s top price has risen over time. Gemini 1.5 Pro cost $1.25 / $5; Gemini 3.1 Pro costs $2 / $12. Gemini 4 Carbon launched at $2 / $10 and doubles to $4 / $20 on December 1, 2026.
GenerationInputOutput
GPT-4 (2023)$30$60
GPT-4 Turbo (2023)$10$30
GPT-4o$2.50$10
GPT-4.1$2$8
GPT-5$1.25$10
GPT-5.2$1.75$14
GPT-5.4$2.50$15
GPT-5.5$5$30
GPT-5.6 Sol$4$20
GPT-6 Sol$2$10
Claude 3 Opus (2024)$15$75
Claude Opus 4 / 4.1$15$75
Claude Opus 4.5 to Opus 5$5$25
Claude Opus 5.5$4$20
Claude 3.5 Sonnet (2024) to Sonnet 4.6$3$15
Claude Sonnet 5 / 5.5$2$10
Gemini 1.5 Pro (2024)$1.25$5
Gemini 2.5 Pro$1.25$10
Gemini 3 Pro / 3.1 Pro$2$12
Gemini 4 Carbon$2 ($4 from Dec 1)$10 ($20 from Dec 1)

Small models: Google got pricier, OpenAI and Anthropic got cheaper

OpenAI mini / Luna: small and fast model list prices by generationGPT-4o mini $0.15 input, $0.60 output; GPT-4.1 mini $0.40 input, $1.60 output; GPT-5 mini $0.25 input, $2 output; GPT-5.4 mini $0.75 input, $4.50 output; GPT-5.6 Luna $0.20 input, $1.20 output; GPT-6 Luna $0.10 input, $0.50 output per 1M tokens.OpenAI mini / LunaInputOutput$0.1$0.3$1$3$10GPT-4o mini: $0.15 input / $0.60 output per 1M tokensGPT-4o mini$0.15 / $0.60GPT-4.1 mini: $0.40 input / $1.60 output per 1M tokensGPT-4.1 miniGPT-5 mini: $0.25 input / $2 output per 1M tokensGPT-5 miniGPT-5.4 mini: $0.75 input / $4.50 output per 1M tokensGPT-5.4 miniGPT-5.6 Luna: $0.20 input / $1.20 output per 1M tokensGPT-5.6 LunaGPT-6 Luna: $0.10 input / $0.50 output per 1M tokensGPT-6 Luna$0.10 / $0.50
Claude Haiku: small and fast model list prices by generationClaude 3 Haiku $0.25 input, $1.25 output; Claude 3.5 Haiku $0.80 input, $4 output; Claude Haiku 4.5 $1 input, $5 output; Claude Haiku 5.5 $0.10 input, $0.50 output per 1M tokens.Claude HaikuInputOutput$0.1$0.3$1$3$10Claude 3 Haiku: $0.25 input / $1.25 output per 1M tokens3 Haiku$0.25 / $1.25Claude 3.5 Haiku: $0.80 input / $4 output per 1M tokens3.5 HaikuClaude Haiku 4.5: $1 input / $5 output per 1M tokensHaiku 4.5Claude Haiku 5.5: $0.10 input / $0.50 output per 1M tokensHaiku 5.5$0.10 / $0.50
Gemini Flash: small and fast model list prices by generationGemini 1.5 Flash $0.075 input, $0.30 output; Gemini 2.0 Flash $0.10 input, $0.40 output; Gemini 2.5 Flash $0.30 input, $2.50 output; Gemini 3 Flash $0.50 input, $3 output; Gemini 3.5 Flash $1.50 input, $9 output; Gemini 3.6 Flash $0.75 input, $3.75 output; Gemini 3.8 Flash $0.75 input, $3.75 output; Gemini 3.8 Flash, Jan 1 $1.50 input, $7.50 output per 1M tokens.Gemini FlashInputOutput$0.1$0.3$1$3$10Gemini 1.5 Flash: $0.075 input / $0.30 output per 1M tokens1.5 Flash$0.075 / $0.30Gemini 2.0 Flash: $0.10 input / $0.40 output per 1M tokens2.0 FlashGemini 2.5 Flash: $0.30 input / $2.50 output per 1M tokens2.5 FlashGemini 3 Flash: $0.50 input / $3 output per 1M tokens3 FlashGemini 3.5 Flash: $1.50 input / $9 output per 1M tokens3.5 FlashGemini 3.6 Flash: $0.75 input / $3.75 output per 1M tokens3.6 FlashGemini 3.8 Flash: $0.75 input / $3.75 output per 1M tokens3.8 FlashGemini 3.8 Flash, Jan 1: $1.50 input / $7.50 output per 1M tokens3.8 Flash, Jan 1$1.50 / $7.50
Official list price per 1M tokens for each lab's small, fast model line. Log scale. Claude 3 Haiku and Gemini 1.5 Flash are launch-era list prices; the last Gemini row is the scheduled 1 January 2027 price.
  • Gemini Flash rose with almost every generation: $0.075 / $0.30 for Gemini 1.5 Flash, $1.50 / $9 for Gemini 3.5 Flash. Gemini 3.6 Flash and 3.8 Flash halved that to $0.75 / $3.75, but both double to $1.50 / $7.50 on January 1, 2027. That is 20 times Gemini 1.5 Flash’s input price.
  • OpenAI’s mini models climbed from $0.15 / $0.60 (GPT-4o mini) to $0.75 / $4.50 (GPT-5.4 mini). The GPT-5.6 and GPT-6 Luna tiers reversed that: GPT-6 Luna costs $0.10 / $0.50, below GPT-4o mini.
  • Claude Haiku went from $0.25 / $1.25 to $1 / $5 with Haiku 4.5. Haiku 5.5 costs $0.10 / $0.50, the cheapest Claude model Anthropic has sold.
GenerationInputOutput
GPT-4o mini$0.15$0.60
GPT-4.1 mini$0.40$1.60
GPT-5 mini$0.25$2
GPT-5.4 mini$0.75$4.50
GPT-5.6 Luna$0.20$1.20
GPT-6 Luna$0.10$0.50
Claude 3 Haiku (2024)$0.25$1.25
Claude 3.5 Haiku$0.80$4
Claude Haiku 4.5$1$5
Claude Haiku 5.5$0.10$0.50
Gemini 1.5 Flash (2024)$0.075$0.30
Gemini 2.0 Flash$0.10$0.40
Gemini 2.5 Flash$0.30$2.50
Gemini 3 Flash$0.50$3
Gemini 3.5 Flash$1.50$9
Gemini 3.6 Flash / 3.8 Flash$0.75 ($1.50 from Jan 1)$3.75 ($7.50 from Jan 1)

The biggest price changes between generations

Change in blended list price from one generation to the nextClaude Haiku 4.5 → 5.5: −90%; GPT-5.4 mini → GPT-5.6 Luna: −73%; Claude Opus 4.1 → 4.5: −67%; Gemini 3.5 Flash → 3.6 Flash: −56%; GPT-5.6 Luna → GPT-6 Luna: −56%; GPT-5.6 Sol → GPT-6 Sol: −50%; Claude Sonnet 4.6 → 5: −33%; Claude Fable 5.1 → 5.2: −20%; Claude Opus 5 → 5.5: −20%; Claude Haiku 3.5 → 4.5: +25%; GPT-5 → GPT-5.2: +40%; GPT-5.4 → GPT-5.5: +100%; Gemini 4 Carbon on 1 Dec 2026: +100%; Gemini 3.8 Flash on 1 Jan 2027: +100%; Gemini 3 Flash → 3.5 Flash: +200%.CheaperPricier−100%−50%0+50%+100%+150%+200%Claude Haiku 4.5 → 5.5: $1/$5 → $0.10/$0.50 per 1M (−90% blended)Claude Haiku 4.5 → 5.5−90%GPT-5.4 mini → GPT-5.6 Luna: $0.75/$4.50 → $0.20/$1.20 per 1M (−73% blended)GPT-5.4 mini → GPT-5.6 Luna−73%Claude Opus 4.1 → 4.5: $15/$75 → $5/$25 per 1M (−67% blended)Claude Opus 4.1 → 4.5−67%Gemini 3.5 Flash → 3.6 Flash: $1.50/$9 → $0.75/$3.75 per 1M (−56% blended)Gemini 3.5 Flash → 3.6 Flash−56%GPT-5.6 Luna → GPT-6 Luna: $0.20/$1.20 → $0.10/$0.50 per 1M (−56% blended)GPT-5.6 Luna → GPT-6 Luna−56%GPT-5.6 Sol → GPT-6 Sol: $4/$20 → $2/$10 per 1M (−50% blended)GPT-5.6 Sol → GPT-6 Sol−50%Claude Sonnet 4.6 → 5: $3/$15 → $2/$10 per 1M (−33% blended)Claude Sonnet 4.6 → 5−33%Claude Fable 5.1 → 5.2: $10/$50 → $8/$40 per 1M (−20% blended)Claude Fable 5.1 → 5.2−20%Claude Opus 5 → 5.5: $5/$25 → $4/$20 per 1M (−20% blended)Claude Opus 5 → 5.5−20%Claude Haiku 3.5 → 4.5: $0.80/$4 → $1/$5 per 1M (+25% blended)Claude Haiku 3.5 → 4.5+25%GPT-5 → GPT-5.2: $1.25/$10 → $1.75/$14 per 1M (+40% blended)GPT-5 → GPT-5.2+40%GPT-5.4 → GPT-5.5: $2.50/$15 → $5/$30 per 1M (+100% blended)GPT-5.4 → GPT-5.5+100%Gemini 4 Carbon on 1 Dec 2026: $2/$10 → $4/$20 per 1M (+100% blended)Gemini 4 Carbon on 1 Dec 2026+100%Gemini 3.8 Flash on 1 Jan 2027: $0.75/$3.75 → $1.50/$7.50 per 1M (+100% blended)Gemini 3.8 Flash on 1 Jan 2027+100%Gemini 3 Flash → 3.5 Flash: $0.50/$3 → $1.50/$9 per 1M (+200% blended)Gemini 3 Flash → 3.5 Flash+200%
Blended price = 3 parts input to 1 part output, a typical chat mix. Each row compares a model with the one it replaced, or a scheduled change to the same model.

The chart compares each model with the one it replaced, using a blended price of 3 parts input to 1 part output. The biggest cuts are recent: Haiku 5.5 is 90% cheaper than Haiku 4.5, and GPT-5.6 Luna is 73% cheaper than GPT-5.4 mini. The biggest rise is Gemini 3 Flash to 3.5 Flash (+200%). Google’s two scheduled changes, Gemini 4 Carbon on December 1 and Gemini 3.8 Flash on January 1, each double the price of a model already on sale.

What this means for your bill

  • A newer model isn’t automatically cheaper. Check the price before you move to a new version. GPT-5.5 cost four times GPT-5’s input price, and Gemini 3.5 Flash cost three times Gemini 3 Flash.
  • Per-token price isn’t cost per task. Reasoning models bill their thinking as output tokens, so a model that thinks longer can cost more even at the same rate. Google pitches Gemini 4 Argon, which costs the same as Carbon, as the model that uses fewer tokens.
  • If you use Gemini, budget for the rises now. From January 1, Gemini 3.8 Flash costs $1.50 / $7.50. Google’s Batch and Flex tiers remain 50% off the standard rate; see our Gemini API pricing guide.
  • The list price isn’t the lowest price. Batch APIs cut 50% at OpenAI, Anthropic and Google, and resellers such as Kie.ai sell GPT, Claude and Gemini models for 30–40% of list price. The cheapest LLM APIs page ranks them, and our Claude API pricing guide covers Anthropic’s discounts.

How we built these charts

  • Prices are official list prices from each lab’s own API (OpenAI, Anthropic, Google), standard real-time tier, shortest context band, in US dollars per 1M tokens.
  • Current models use the prices in our listings on November 17, 2026. Claude Haiku 5.5, Claude Fable 5.2 and Gemini 4 Carbon aren’t in our listings yet; their prices come from Anthropic’s and Google’s pricing pages on the same date.
  • Retired models (GPT-4, GPT-4 Turbo, Claude 3 Opus, Claude 3.5 Sonnet, Claude 3 Haiku, Gemini 1.5 Pro and Gemini 1.5 Flash) use their published list prices while on sale. For Gemini 1.5 we use the prices after Google’s 2024 price cut.
  • GPT-4o shows its later $2.50 / $10 price; it launched at $5 / $15.
  • The price axis is logarithmic, so equal distances mean equal percentage changes.

You can reuse these charts and tables with a link to this page. We’ll update them as prices change.

Subscribe to our newsletter

News and updates, 1–2 emails per month. No spam.