AI Tools
ChatGPT vs Claude: which AI assistant is the better buy?
For the ChatGPT vs Claude decision, verified API pricing as of August 10, 2026 favors Anthropic: Sonnet 5 costs $2 per million input and $10 per million output tokens through August 31, 2026, undercutting OpenAI's comparable gpt-5.6-terra at $2/$12. Both vendors publish current model prices, and the gap between the two flagship families is smaller than the marketing suggests.
What matters
- Claude's Sonnet 5 API at $2 in / $10 out per million tokens beats OpenAI's gpt-5.6-terra ($2 / $12) until August 31, 2026, then rises to $3 / $15.
- On the budget tier OpenAI wins: gpt-5.6-luna at $0.20 / $1.20 per million tokens is the cheapest flagship-family model on either platform.
- Anthropic publishes consumer plan prices openly (Free $0, Pro $17/mo annual or $20 monthly, Max from $100/mo); OpenAI's ChatGPT pricing page blocks automated access.
- Both vendors sell premium throughput: OpenAI's Fast mode (renamed from Priority on July 30, 2026) doubles standard prices; Anthropic's Opus 5 fast mode costs 2x standard.
What are ChatGPT and Claude?
ChatGPT is OpenAI's AI assistant product, powered by the gpt-5.x model family, and sold as both a subscription app and an API. Claude is Anthropic's assistant, powered by the Claude 5 family (Fable 5, Opus 5, Sonnet 5, Haiku 4.5), and sold the same way: consumer subscriptions and a developer API.
An analysis shared on Hacker News this week estimates that OpenAI and Anthropic together capture roughly 70% of AI industry revenue, which is why the ChatGPT vs Claude choice dominates so many buying decisions. For most buyers the practical differences come down to three things: API cost per token, what the consumer plans include, and how each model is positioned for coding and agent work.
ChatGPT vs Claude API pricing compared
Both vendors publish per-token prices for current models. The table below uses the standard tier, short context, per 1 million tokens, as verified on the official pricing pages on August 10, 2026.
Three patterns stand out. First, OpenAI's budget model gpt-5.6-luna at $0.20 / $1.20 is dramatically cheaper than Claude's cheapest current model, Haiku 4.5 at $1 / $5. Second, at the mid tier Sonnet 5 is the better deal while its introductory pricing lasts. Third, at the flagship tier the two are nearly identical, with Claude's Opus 5 at $5 / $25 slightly undercutting OpenAI's gpt-5.6-sol at $5 / $30.
Batch pricing cuts both vendors' standard rates by 50%. OpenAI also offers Fast mode, renamed from Priority processing on July 30, 2026, at roughly 2x standard rates, and Anthropic sells fast mode for Opus 5 at 2x standard.
Consumer plans: what each subscription includes
Anthropic's pricing page lists three consumer tiers: a free plan at $0, Pro at $17 per month with annual billing ($200 billed up front) or $20 billed monthly, and Max from $100 per month with 5x or 20x usage options. Pro adds Claude Code, Claude Cowork, and access to more Claude models on top of the free tier.
OpenAI does not publish ChatGPT consumer plan pricing on a page reachable by automated checks, so this comparison cannot verify exact ChatGPT subscription prices. The verified alternative is the API model list: the ChatGPT product itself is available as a paid subscription, and the chat-latest API model is priced at $5 / $30 per million tokens. Check openai.com in a browser for current consumer plan details.
Coding and agentic work: where the models differ
Anthropic positions its flagship Opus 5 for complex agentic coding and enterprise work at $5 / $25 per million tokens, and its newest Fable 5 for long-running agents at $10 / $50. Claude Code is included in the Pro and Max consumer plans rather than sold as a separate API SKU.
OpenAI prices a dedicated coding model, gpt-5.3-codex, at $1.75 in / $14 out per million tokens on the standard tier, and sells Codex as a product alongside ChatGPT. For heavy coding workloads, the verified difference is structural: Claude bundles its coding tool into subscriptions, while OpenAI prices codex separately, so the cheaper route depends on whether you pay per token or per seat.
The honest catch buyers miss
Three things are easy to miss when comparing these two.
- Sonnet 5's price is temporary. The $2 / $10 rate runs only through August 31, 2026, then becomes $3 / $15 per million tokens. Budgets built on the introductory rate will overshoot by 50% on output tokens.
- OpenAI's consumer pricing is opaque. Its ChatGPT pricing page returns a bot check to curl and to headless browsers, so plan prices cannot be verified from a public page. Anthropic publishes everything.
- Regional and priority surcharges exist on both sides. OpenAI charges a 10% uplift on data-residency endpoints for models released on or after March 5, 2026. Anthropic charges 1.1x for US-only inference, and its prompt caching prices assume a 5-minute TTL.
Which should you choose?
Pick ChatGPT when raw API cost is the constraint and your workload fits a small model, because gpt-5.6-luna has no Claude equivalent near its price. Pick Claude when you want mid-tier quality today, because Sonnet 5's introductory $2 / $10 rate undercuts gpt-5.6-terra until August 31, 2026, or when you want a single subscription that bundles Claude Code.
Pick either flagship tier based on model quality rather than price, since Opus 5 and gpt-5.6-sol differ by only $5 per million output tokens. And if your team already standardizes on one ecosystem, staying there usually beats switching for a per-token saving of a few dollars.
At a glance
| Model | Vendor | Input | Output | Notes |
|---|---|---|---|---|
| gpt-5.6-sol | OpenAI | $5.00 | $30.00 | Flagship; cached input $0.50 |
| gpt-5.6-terra | OpenAI | $2.00 | $12.00 | Mid tier; cached input $0.20 |
| gpt-5.6-luna | OpenAI | $0.20 | $1.20 | Budget; cheapest current flagship-family model |
| Fable 5 | Anthropic | $10.00 | $50.00 | Newest model, built for long-running agents |
| Opus 5 | Anthropic | $5.00 | $25.00 | Flagship; positioned for complex agentic coding |
| Sonnet 5 | Anthropic | $2.00 | $10.00 | Intro price to Aug 31, 2026, then $3 / $15 |
| Haiku 4.5 | Anthropic | $1.00 | $5.00 | Fastest and most cost-efficient Claude model |
FAQ
Which is cheaper for API use, ChatGPT or Claude?
It depends on the tier. OpenAI's gpt-5.6-luna at $0.20 / $1.20 per million tokens is the cheapest model on either platform, but Claude's Sonnet 5 at $2 / $10 undercuts the comparable gpt-5.6-terra at $2 / $12 until August 31, 2026, when Sonnet 5 rises to $3 / $15.
Is Claude cheaper than ChatGPT for a subscription?
Anthropic publishes its plans: Free at $0, Pro at $17 per month with annual billing or $20 monthly, and Max from $100 per month. OpenAI's ChatGPT pricing page blocks automated access, so its current consumer plan prices could not be verified for this article; check openai.com in a browser.
Is Claude better than ChatGPT for coding?
Anthropic positions Opus 5 for complex agentic coding and bundles Claude Code into its Pro and Max subscriptions. OpenAI prices a dedicated gpt-5.3-codex model at $1.75 / $14 per million tokens. The cheaper option depends on whether you pay per token or per subscription seat.
Related reading
Claude tool profile, GPT-5.6 radar, GPT-5.6 Sol vs Luna comparison, Radar index