ChatGPT Pro 500 Adds GPT-6 Astra Ultrafast
ChatGPT Pro 500 is a new $500 monthly plan with the highest included Pro usage and access to GPT-6 Astra Ultrafast in ChatGPT Work and Codex, followed by credit-based usage after the included allowance is spent.
ChatGPT Pro 500 is a new $500-per-month subscription that gives individual Pro users the highest included usage among the Pro plans and unlocks GPT-6 Astra Ultrafast in ChatGPT Work and Codex. OpenAI says Ultrafast uses the plan’s included allowance first and then draws from the user’s credit balance. Pro 100 and Pro 200 users cannot unlock Ultrafast merely by buying credits, while Enterprise users also receive access. Pro 500 subscriptions are available through ChatGPT on the web at launch. Read the September 29 ChatGPT release notes.
OpenAI separately launched GPT-6 Astra Ultrafast in the API. The company reports up to 8x faster token generation in Codex, with a cited rate of 300 tokens per second, and up to 6x in the API. Those figures describe the premium service tier’s generation speed, not a blanket reduction in task completion time. Tool calls, prompt processing, network latency and human review can still dominate an end-to-end workflow. See OpenAI’s DevDay 2026 recap.
What ChatGPT Pro 500 includes
Pro 500 sits above Pro 100 and Pro 200 as a distinct subscription, not as a credit bundle added to a lower plan. Its published monthly price is $500. OpenAI calls it the Pro plan with the highest included usage; the DevDay recap describes that allowance as 25 times the ChatGPT Plus allowance. The release notes do not publish a token count or a fixed number of Astra Ultrafast tasks, so the multiplier should not be converted into an invented request quota.
The plan’s defining access change is the Ultrafast option for GPT-6 Astra. In a supported product, a Pro 500 subscriber selects Ultrafast in the model picker and receives the faster service tier. The named surfaces are ChatGPT Work and Codex. OpenAI does not say this launch places Ultrafast in an ordinary Chat conversation, so availability should be checked on the exact surface rather than inferred from the ChatGPT subscription alone.
Enterprise is the other plan named for ChatGPT Work and Codex access in the DevDay announcement. For an individual buyer, however, the new route is specifically Pro 500. The site’s GPT-6 Astra model page remains the reference for the model’s capabilities; this release changes the available speed tier and subscription route rather than announcing a new Astra model ID.
How the included allowance and credits work
OpenAI describes a sequence: Astra Ultrafast first consumes the included Pro 500 allowance, then uses the account’s credit balance after that allowance is exhausted. That makes three different charges important to separate. The $500 amount is the recurring subscription price. The included allowance is usage packaged with that subscription. Credits are an additional balance used after the allowance for this product experience.
Credits do not function as a universal access key. OpenAI explicitly says that buying credits on Pro 100 or Pro 200 does not unlock Ultrafast. A user who needs this tier therefore cannot reproduce Pro 500 access by keeping a lower subscription and purchasing more credits. The plan entitlement must exist before the balance can extend usage.
The official sources captured for this announcement do not state the size of the included allowance in tokens, the credit conversion rate for a particular workload, or the cost of one Ultrafast task. A cost comparison should use the account’s displayed allowance and credit terms rather than deriving them from API token prices. The access guide can help distinguish product plan availability from model availability.
Astra Ultrafast pricing is separate from API billing
The API launch is a separate access path. OpenAI says GPT-6 Astra Ultrafast is available in the API today, but the two announcement sources used here do not provide its API token rates. The $500 ChatGPT Pro 500 subscription price is therefore not an API price, and a Pro 500 allowance should not be entered into an API cost model as if it were prepaid API credit. Conversely, API access does not prove that a ChatGPT account has the Pro 500 product entitlement.
This distinction matters for teams comparing routes. A Pro 500 user can choose Ultrafast inside the supported ChatGPT Work or Codex experience, with included allowance followed by credits. An API developer selects the available API service tier and is billed under the API’s published terms. Enterprise customers have product access named in the announcement, but their commercial terms may be governed by an organization agreement rather than the individual Pro 500 price.
Until OpenAI publishes or verifies the applicable Ultrafast API rate in its pricing documentation, existing evergreen prices and calculators should not be silently relabeled. The API pricing guide covers the established token-billing model, and the model comparison tool keeps model capabilities separate from plan entitlements and speed tiers.
What the 8x and 6x speed claims measure
OpenAI’s stated comparison is up to 8x faster token generation in Codex and up to 6x in the API. The Codex claim also names 300 tokens per second. The phrase “up to” matters because it describes a ceiling observed by the provider, not a minimum for every prompt, region, context length or load condition. It also concerns generated tokens rather than total workflow duration.
For a coding task, elapsed time can include repository setup, retrieval, shell commands, tests and code review. For an agent workflow, tool execution may take longer than model generation. A short answer may finish quickly enough that startup latency matters more than sustained tokens per second. Ultrafast is most likely to matter when output generation is a material share of the critical path, especially in interactive coding or long-form responses where waiting for tokens slows the next action.
A useful evaluation records time to first token, sustained generation rate, total task time, tool time, retries and reviewer corrections. Compare the same GPT-6 Astra task on the standard and Ultrafast tiers with the same inputs and acceptance criteria. That test can show whether the premium tier improves the workflow that matters, without converting a provider generation-speed figure into an unsupported claim about productivity or quality.
How to verify access before changing plans
A prospective subscriber should first confirm that the intended work happens in ChatGPT Work or Codex, the two product surfaces named for Pro 500. After subscribing on the web, check that GPT-6 Astra is selected and that Ultrafast appears as a service-tier choice. Record whether a test run consumes included allowance or credits, because the billing stage changes after the allowance is used.
Teams should also record the account plan, workspace type, product surface, region, selected model and service tier. For API tests, record the endpoint, model ID, processing tier and billed usage separately. This prevents a missing product entitlement from being misdiagnosed as an API problem, or an API billing line from being mistaken for ChatGPT credits.
ChatGPT Pro 500’s practical offer is precise: a $500 monthly Pro subscription, the largest included allowance in the Pro range, and GPT-6 Astra Ultrafast access in ChatGPT Work and Codex before usage continues against credits. Its value depends on whether faster token generation shortens real tasks enough to justify the higher plan and any additional credits—not on treating an “up to” speed number as a guarantee.