# Sonnet 5.5, GPT-6.1 Sol and Argon share a price tag, not a bill

> GPT-6.1 Sol vs Sonnet 5.5 pricing looks identical at $2 and $10 per million tokens, but one roundup puts the cost per task anywhere from $0.72 to $7.60.

- Published: 2026-10-12
- Topic: [AI](https://futuretechnologyhq.com/topic/ai/)
- URL: https://futuretechnologyhq.com/article/sonnet-5-5-gpt-6-1-sol-gemini-4-argon-cost-per-task/
- Publisher: [Future Technology](https://futuretechnologyhq.com/), free to read. Quote as "Future Technology" with a link to the article.

## Key points

- Sonnet 5.5, GPT-6.1 Sol and Gemini 4 Argon all list at $2 input and $10 output per million tokens, but a roundup puts their cost per index task at $1.08 to $7.60, $0.72 and $1.99.
- Sonnet 5.5 costs about seven times more per task at max effort ($7.60) than at high effort ($1.08), and the API defaults to high.
- Argon's $2 and $10 rate is introductory, with $4 and $20 as the standard price, and access is limited to trusted cyber defenders.

## Article

*Last updated: 12 October 2026*

Three frontier models list at the same $2 per million input tokens and $10 per million output tokens. According to one October roundup, the cost of finishing a standard benchmark task on them ranges from $0.72 to $7.60.

That roundup comes from Local AI Zone, a single aggregator whose page carries inconsistent dates. The $2 and $10 list prices match Anthropic's and OpenAI's own pricing pages, but treat the per-task figures below as reported rather than independently checked.

## What the roundup reports

Claude Sonnet 5.5 scores 56.0 on the Intelligence Index at max effort, at $7.60 per task. At xhigh effort it scores 51.9 for $2.74, and at high effort it scores 46.7 for $1.08. The API defaults to high effort, so the default score is 46.7, not the 56.0 headline. At max effort, one index run used 410 million output tokens.

GPT-6.1 Sol scores 51.8 at $0.72 per task, though the roundup does not state the effort level. Above 272K prompt tokens, every GPT-6 tier re-bills the whole request at twice the input price and 1.5 times the output price. OpenAI's pricing page lists Sol's long context rate at $4 input and $15 output. The roundup says Sol is not yet in ChatGPT's chat window.

Gemini 4 Argon scores 52.6 at high effort (The Batch puts it at 53) and costs $1.99 per task at its introductory $2 and $10 pricing. The standard price is $4 and $20, so that figure will rise. Access is limited to trusted cyber defenders through the Fairwind Program, and the roundup says no general availability date was given and Google had not confirmed one as of 7 October.

## Why the sticker price misleads

Your bill is the price per token multiplied by the tokens a model spends, and effort level changes how many it spends. Sonnet 5.5 shows it clearly: the same list price produces a cost per task about seven times higher at max effort than at high.

The cheapest figure, Sol at $0.72, also needs care. Its effort level is unstated, so it is not a like-for-like match with Sonnet's rows. Argon looks cheap only while the introductory rate lasts.

For more on how these models compare, our earlier pieces cover [the GPT-6 Astra launch](/openai-gpt-6-astra-launch/) and [Claude Sonnet 5.5 pricing](/article/claude-sonnet-5-5-pricing/). If per-token bills worry you, our [local LLM VRAM guide](/article/local-llm-vram-tier-list-2026/) shows what it takes to run models yourself.

The numbers to watch are Argon's general availability date and its standard price, plus whether independent testers reproduce these per-task costs.

## Sources

- [October 2026 AI model updates](https://local-ai-zone.github.io/blog/October_2026_AI_Model_Updates.html)
- [Claude pricing (Anthropic)](https://claude.com/pricing)
- [API pricing (OpenAI)](https://developers.openai.com/api/docs/pricing)
- [DeepLearning.AI)](https://www.deeplearning.ai/the-batch/argon-a-low-hallucination-model-build-for-knowledge-work)