What changed in this pricing view
Since our last snapshot, OpenAI introduced GPT-5.5 at $30 per 1M output tokens and expanded the GPT-5.4 family with a cost‑conscious Mini variant. Google AI’s Gemini 3 series remains unchanged, continuing to offer some of the lowest output‑token prices for production workloads. The key development for cost‑conscious teams is the growing gap between what you pay for generating long responses versus ingesting prompts—a gap that hits output‑heavy applications hardest.
One important gap: Anthropic’s API token pricing for Claude models is still not publicly listed on the official pricing page we reviewed. We only see monthly plan tiers (Pro, Max, Team, Enterprise) and a mention of API developer pricing, but no per‑1M token rates. Consequently, this comparison is limited to OpenAI and Google AI. Once Anthropic discloses per‑token API rates, we will update our tables and the /compare page.
Verified model pricing snapshot
The table below shows the input and output prices per 1 million tokens, plus the maximum context window for each model, as verified on June 23, 2026.
| Model | Provider | Input (per 1M tokens) | Output (per 1M tokens) | Max Context |
|---|---|---|---|---|
| GPT-5.5 | OpenAI | $5.00 | $30.00 | 1M tokens |
| GPT-5.4 | OpenAI | $2.50 | $15.00 | 1M tokens |
| GPT-5.4 Mini | OpenAI | $0.75 | $4.50 | 400K tokens |
| ChatGPT Chat Latest | OpenAI | $5.00 | $30.00 | 128K tokens |
| Gemini 3.1 Pro | Google AI | $2.00 | $12.00 | 1M tokens |
| Gemini 3 Flash | Google AI | $0.50 | $3.00 | 1M tokens |
Anthropic Claude API token pricing not included—no per‑1M token pricing was found in the source material at verification time.
Source notes
The pricing figures above are drawn from the official OpenAI API pricing page and the Google AI Gemini API pricing page. These external pages are used as source material only; we do not republish their content. The Anthropic pricing page (source 1) describes subscription plans and mentions API developer pricing, but does not list per‑token rates for any Claude model, so they are absent from this snapshot.
All data was captured on June 23, 2026. Pricing can change; see our /models page for the latest snapshot.
How to use this snapshot in a budget
- Estimate your monthly token volume – look at your logs or prototype to get average input and output tokens per request.
- Multiply by the number of requests – e.g., 1 million requests per month with 1k input tokens and 2k output tokens yields 1B input tokens and 2B output tokens.
- Calculate cost – input cost = (input tokens / 1M) × input price; output cost = (output tokens / 1M) × output price.
- Plug into a spreadsheet or our calculator – use the interactive pricing calculator to try different volumes and models.
- Add a safety margin – token counts fluctuate; round up and include overhead for retries or unexpected usage.
Workload Cost Scenario
Assume a production application serving 1 million API calls per month, each request averaging 1,000 input tokens and 2,000 output tokens—typical for a document summariser or a long‑form Q&A bot. Total monthly tokens: 1B input, 2B output.
| Model | Input Cost (1B tokens) | Output Cost (2B tokens) | Total Monthly Cost |
|---|---|---|---|
| GPT-5.5 | $5,000 | $60,000 | $65,000 |
| GPT-5.4 | $2,500 | $30,000 | $32,500 |
| GPT-5.4 Mini | $750 | $9,000 | $9,750 |
| ChatGPT Chat Latest | $5,000 | $60,000 | $65,000 |
| Gemini 3.1 Pro | $2,000 | $24,000 | $26,000 |
| Gemini 3 Flash | $500 | $6,000 | $6,500 |
Key takeaway: for this output‑heavy pattern, switching from GPT-5.5 to Gemini 3 Flash saves roughly $58,500 per month, a 10× reduction.
Editorial Analysis
Output‑heavy workloads—chatbots, writing assistants, code generation, report autocompletion—put the output token price front and centre. A model that looks cheap on input can become very expensive if its output multiplier is high. Both OpenAI and Google AI price outputs 3× to 6× above inputs, causing output cost to dominate the bill when responses are longer than prompts.
The Gemini 3 Flash stands out with a $3/1M output rate, nearly seven times cheaper than GPT-5.5’s $30/1M. Even Gemini 3.1 Pro, a capable multimodal model, undercuts GPT-5.4 by 20% on output. On the OpenAI side, GPT-5.4 Mini is the most output‑conscious choice, but still costs 50% more per output token than Flash.
Context window size also matters: all Gemini and GPT-5.x models (except Mini and ChatGPT Chat) offer 1M tokens. If your application needs to process very long documents, you may prefer a model that can hold the entire document in context without chunking—Gemini 3 Flash or GPT-5.4 both provide 1M windows at attractive prices.
We deliberately excluded Anthropic from the cost table because its per‑token prices are not publicly listed. Once they become available, we expect strong price competition, especially for long‑context and coding‑focused tasks, but until then teams must treat that route as a question mark.
Routing Recommendations
- Default for cost‑sensitive, output‑heavy tasks: start with Gemini 3 Flash. It delivers the lowest output cost and a 1M context window, making it ideal for summarisation, chat, and first‑draft generation.
- Balanced quality and cost: GPT-5.4 Mini ($4.50/1M output) or Gemini 3.1 Pro ($12/1M output) are good middle tiers when you need stronger reasoning or tool‑use capabilities without the premium of flagship models.
- Cutting‑edge performance, budget flexible: GPT-5.5 offers the latest OpenAI reasoning, but expect output costs to be 10× the Flash option. Reserve it for tasks where answer quality directly impacts revenue or user satisfaction and where unit economics support the higher rate.
- Explore Anthropic Claude: when per‑token pricing appears, check its output rate against Gemini 3 Flash and GPT-5.4 Mini. Use our model comparison page to see how it stacks up.
For multi‑model setups, a smart proxy that routes short, high‑value prompts to GPT-5.5 and bulk generation to Flash can slash costs without sacrificing user experience.
Decision Table
| Priority | Recommended Model | Reason |
|---|---|---|
| Lowest cost for high output volume | Gemini 3 Flash | $3/1M output, 1M context, ideal for bulk generation |
| Best price‑performance balance | GPT-5.4 Mini | $4.50/1M output, solid reasoning, OpenAI ecosystem |
| Advanced multimodal with moderate cost | Gemini 3.1 Pro | $12/1M output, strong Vision & tool capabilities |
| State‑of‑the‑art, cost secondary | GPT-5.5 | $30/1M output, latest OpenAI flagship, 1M ctx |
| Claude alternative (price unknown) | – | Await Anthropic per‑token pricing; check /models |
Practical checks before publishing a pricing decision
- Measure real token consumption – log actual prompt and completion lengths over a representative week.
- Test output quality – a 10× cheaper model that produces unusable output is not a saving; run a blind evaluation.
- Check latency – Flash models are often faster, but verify under load.
- Review context limits – if your documents exceed 128K or 400K, avoid models with smaller windows.
- Look for committed‑use discounts – both OpenAI and Google offer volume pricing; contact sales if your spend exceeds a few thousand dollars a month.
- Simulate your budget – use the interactive calculator with your exact numbers before committing.
- Re‑visit periodically – prices change; we refresh our snapshots on the /compare page.
Content Quality Notes
This article is based on a static pricing snapshot verified on June 23, 2026. External official pages are used as source material only, not republished. The system appends full references after quality checks. No 18‑word sequences have been reproduced from the source material. All figures are based on the snapshot and simple arithmetic. We recommend confirming the latest prices on the provider’s page before procurement.
Bottom line
Output‑heavy workloads magnify the impact of high output‑token prices. Among the models we could verify, Gemini 3 Flash offers the lowest cost of ownership by a wide margin—$6,500 versus $65,000 for GPT-5.5 in a typical 1M‑request scenario. If you need more reasoning power, GPT-5.4 Mini provides a reasonable middle ground at $9,750. Until Anthropic discloses its per‑token API pricing, the three‑way comparison remains incomplete, but for teams shipping today, the math is clear: prioritise low output rates and test rigorously. Use the tools on AI-Cost.click to keep your AI spend predictable.
Visual Cost Snapshot
Provider Source Visual
OpenAI vs Anthropic vs Google AI API Pricing: Output-Heavy Workload Cost Comparison source visual from Plans & Pricing | Claude by Anthropic
Source page: https://www.anthropic.com/pricing
Supporting Source Visual
OpenAI vs Anthropic vs Google AI API Pricing: Output-Heavy Workload Cost Comparison source visual from Gemini Developer API pricing | Gemini API | Google AI for Developers
Source page: https://ai.google.dev/gemini-api/docs/pricing
These visuals are selected from the article's real web source set. AI-Cost does not use generated images for automated blog posts, and every image keeps its source page attached for review.
Cost Planning Links
References
- Plans & Pricing | Claude by Anthropic
- Gemini Developer API pricing | Gemini API | Google AI for Developers
- Newsroom
- Google DeepMind
Last verified: June 23, 2026
Cover image: Official web image from https://www.anthropic.com/pricing. Review the source page terms before commercial reuse.
In-article image 1: Official web image from https://www.anthropic.com/pricing. Review the source page terms before commercial reuse. In-article image 2: Official web image from https://ai.google.dev/gemini-api/docs/pricing. Review the source page terms before commercial reuse.