GPT-6.1 Sol Ultrafast: price list and multipliers
As of 2026-10-13, the Ultrafast row in OpenAI's official price list reads gpt-6.1-sol $12.00 $0.60 $15.00 $60.00 $24.00 $1.20 $30.00 $90.00: short context $12.00 input, $0.60 cached input, $15.00 cache write and $60.00 output, long context $24.00, $1.20, $30.00 and $90.00, all per million tokens. The Standard row in the same table is gpt-6.1-sol $2.00 $0.10 $2.50 $10.00 $4.00 $0.20 $5.00 $15.00, so Ultrafast costs 6 times that row when converted from the official list. The other multiplier is on the subscription side: Ultrafast uses included subscription limits at 8x the Standard rate., while purchased credits and Enterprise pay-as-you-go usage are billed at 6x the Standard rate. The tier entered the Responses API on 2026-10-08, switched on with service_tier set to ultrafast.
Updated 2026-10-13
Prices change: on QCode, Claude, GPT and Chinese models are billed per token — live rates on /models.
- Billed per token — live rates on /models
- Pay by card (Visa / Mastercard / AMEX), Apple Pay, Google Pay or crypto
- Self-service: top up and activate instantly
Four numbers
Input / output (short context, per 1M tokens)
The Ultrafast row in the official price list reads gpt-6.1-sol $12.00 $0.60 $15.00 $60.00 $24.00 $1.20 $30.00 $90.00. The first four numbers are short-context input, cached input, cache write and output.
Converted from the official price list, vs Standard
The Standard row in the same table is gpt-6.1-sol $2.00 $0.10 $2.50 $10.00 $4.00 $0.20 $5.00 $15.00: $12.00 against $2.00 and $60.00 against $10.00, so all four unit prices are 6 times. That multiple is our arithmetic on the official list, and OpenAI's own gpt-6.1-sol model page states Ultrafast mode prices are 6x Standard., which matches it.
Codex and ChatGPT Work multipliers
Two official sentences: Ultrafast uses included subscription limits at 8x the Standard rate. and Ultrafast mode bills purchased credits and Enterprise pay-as-you-go usage at 6x the Standard rate. Both describe how subscription usage is drawn down, not what the API charges per token.
Default Ultrafast TPM (Build / Launch / Grow)
The official defaults for gpt-6.1-sol are Build 1,000,000, Launch 4,000,000 and Grow 40,000,000, and the docs state that Ultrafast has separate rate limits from Standard and Fast modes.
What the tier is, and when GPT-6.1 Sol got it
As of 2026-10-13, the official guide states Ultrafast mode is the fastest service tier in the OpenAI API. It is broadly available for GPT-6 Astra and GPT-6.1 Sol. The setup line is Set model to gpt-6-astra or gpt-6.1-sol and service_tier to ultrafast in each response.create event. For GPT-6.1 Sol the API changelog entry dated 2026-10-08 reads Added Ultrafast mode for GPT-6.1 Sol in the Responses API. and adds It is available to all API users, subject to rate limits, with global processing and US and EU data residency. The same day's ChatGPT & Codex changelog is headed GPT-6.1 Sol Ultrafast in Codex and ChatGPT Work. As for how much faster, the only official number is about Astra: GPT-6 Astra Ultrafast generates tokens up to 8x faster than GPT-6 Astra in Standard mode in Codex. which is the vendor's own measurement; the API changelog, the Ultrafast guide and the Codex speed page checked on 2026-10-13 carry no speed figure for GPT-6.1 Sol Ultrafast.
The vendor's wording (two entries on 2026-10-08)
API side, verbatim: Added Ultrafast mode for GPT-6.1 Sol in the Responses API. Use gpt-6.1-sol with service_tier: "ultrafast" to reduce the time between generated output tokens. The ChatGPT & Codex changelog for the same date says Ultrafast mode speeds up token generation with GPT-6.1 Sol in Codex and ChatGPT Work. and then Ultrafast is available on Pro $500 and eligible Enterprise and Edu plans. Enterprise access is off by default; workspace owners must enable it. The help-center article 20001516 puts the scope harder: Other self-serve plans, including Plus, Pro $100, Pro $200, and Business, do not include Ultrafast at launch. Purchasing credits on one of these plans does not enable it. (the wording here comes from a web.archive.org copy of that page). Where the vendor talks about speed it writes reduce the time between generated output tokens, and up to 8x faster for Astra; both are vendor statements.
Ultrafast and GPT-6.1 Sol, in dates
The vendor releases the model: Released GPT-6.1 Sol ( gpt-6.1-sol ) for complex coding and professional work at a lower cost than GPT-6 Astra. The same entry gives the baseline: Standard pricing per 1M tokens for prompts with up to 272K input tokens is $2 input, $0.10 cached input, $2.50 cache write, and $10 output. Codex and ChatGPT Work take the model the same day (GPT-6.1 Sol in Codex and ChatGPT Work). At that point Ultrafast exists only on the other model: Added Ultrafast mode for GPT-6 Astra in the Responses API.
The API changelog adds Ultrafast to gpt-6.1-sol: Added Ultrafast mode for GPT-6.1 Sol in the Responses API. The ChatGPT & Codex changelog publishes GPT-6.1 Sol Ultrafast in Codex and ChatGPT Work the same day and notes that Enterprise access is off by default until workspace owners enable it.
Checked on 2026-10-13: on the official price list the row still reads gpt-6.1-sol $12.00 $0.60 $15.00 $60.00 $24.00 $1.20 $30.00 $90.00 and the default still reads Build 1,000,000 TPM. The two multipliers are still 6x by conversion of the price list and 8x in the Codex docs, written separately by the vendor.
Confirmed vs the two numbers that diverge
Confirmed (verbatim in the official pages)
All of this is vendor wording on official pages checked on 2026-10-13: Added Ultrafast mode for GPT-6.1 Sol in the Responses API. and Use gpt-6.1-sol with service_tier: "ultrafast" to reduce the time between generated output tokens. the Ultrafast price row gpt-6.1-sol $12.00 $0.60 $15.00 $60.00 $24.00 $1.20 $30.00 $90.00, the Standard row gpt-6.1-sol $2.00 $0.10 $2.50 $10.00 $4.00 $0.20 $5.00 $15.00 and the gpt-6-astra row gpt-6-astra $60.00 $6.00 $75.00 $300.00 $120.00 $12.00 $150.00 $450.00; the context boundary Short context: ≤272K input tokens. Long context: >272K input tokens.; the gpt-6.1-sol model page adds Prompts with more than 272K input tokens are priced at 2x input and cache rates and 1.5x output for the full request. and Ultrafast mode prices are 6x Standard.; the default Ultrafast limits, headed GPT-6.1 Sol has the following default Ultrafast token rate limits: Build 1,000,000, Launch 4,000,000 and Grow 40,000,000 for gpt-6.1-sol and Build 500,000, Launch 1,000,000, Grow 5,000,000 for gpt-6-astra; on the subscription side Ultrafast uses included subscription limits at 8x the Standard rate. and 6x the Standard rate; plan scope Ultrafast is available in Codex and ChatGPT Work on Pro $500 and eligible Enterprise and Edu plans; For Enterprise workspaces, Ultrafast is off by default; and Other self-serve plans, including Plus, Pro $100, Pro $200, and Business, do not include Ultrafast at launch. Purchasing credits on one of these plans does not enable it. Two things here are not vendor wording: converting the two price rows into a how-many-times figure (though the model page does state Ultrafast mode prices are 6x Standard.), and putting the API and subscription multipliers side by side.
Unverified, and absent from the pages checked here
Two official numbers do not line up, and this page does not reconcile them. Converted from the official price list, all four unit prices are 6 times Standard (input $2.00 becomes $12.00, output $10.00 becomes $60.00), and the gpt-6.1-sol model page states Ultrafast mode prices are 6x Standard., while the Codex and ChatGPT Work docs state Ultrafast uses included subscription limits at 8x the Standard rate. and bill purchased credits and Enterprise pay-as-you-go usage at 6x. The price list, the model page, the Ultrafast guide and the Codex docs checked on 2026-10-13 give no reason why the API unit price is 6 times while subscription draw-down is 8x, and no bill type for either number. Those pages carry no speed figure for GPT-6.1 Sol Ultrafast: the only number, GPT-6 Astra Ultrafast generates tokens up to 8x faster than GPT-6 Astra in Standard mode in Codex., is about Astra and is the vendor's own testing. Those pages also list no Ultrafast-specific context window or maximum output, no latency or throughput range, and no effective or review date for these prices.
Against Standard and against Fast
vs Standard (same gpt-6.1-sol)
Standard reads gpt-6.1-sol $2.00 $0.10 $2.50 $10.00 $4.00 $0.20 $5.00 $15.00 and Ultrafast reads gpt-6.1-sol $12.00 $0.60 $15.00 $60.00 $24.00 $1.20 $30.00 $90.00, i.e. 6 times on every unit price when converted from the official list. The limits are counted separately: Ultrafast has separate rate limits from Standard and Fast modes. Whether that is worth paying for is left to you, and the guide gives one line for it: Use it when speed justifies the higher cost.
vs Fast, and vs gpt-6-astra
The Fast row for the same model is gpt-6.1-sol $4.00 $0.20 $5.00 $20.00 $8.00 $0.40 $10.00 $30.00, i.e. 2 times Standard by conversion, and Codex writes 2.5x the Standard rate for included subscription limits and 2x for purchased credits and Enterprise pay-as-you-go, against 8x and 6x for Ultrafast. gpt-6-astra Ultrafast costs $60.00, $6.00, $75.00 and $300.00 per million tokens in short context, with default limits of Build 500,000, Launch 1,000,000 and Grow 5,000,000 TPM, and the vendor says it supports US data residency and global processing only; gpt-6.1-sol adds EU residency.
How to use these numbers together
First, identify the tier: Ultrafast mode is the fastest service tier in the OpenAI API. Configure it as the docs write: Set model to gpt-6-astra or gpt-6.1-sol and service_tier to ultrafast in each response.create event. Second, find your context band: the price list states Short context: ≤272K input tokens. Long context: >272K input tokens., and the model page states Prompts with more than 272K input tokens are priced at 2x input and cache rates and 1.5x output for the full request., so past 272K of input the whole request uses the long-context numbers, $24.00, $1.20, $30.00 and $90.00 for gpt-6.1-sol. Third, connection style has an official preference: We strongly recommend WebSockets , especially for agentic applications that make many tool calls in quick succession. Without a persistent connection, network overhead can reduce the latency gains. Fourth, check limits before raising traffic: the guide says Check your organization’s limits before increasing traffic. and the defaults for gpt-6.1-sol are Build 1,000,000, Launch 4,000,000 and Grow 40,000,000 TPM, with the organisation tier upgrading automatically as total credit purchases reach each threshold (Build $5 in total credit purchases $500 / month at entry, Grow $500 in total credit purchases $200,000 / month at the top). Codex with an API key is a different bill: With an API key, Codex uses API token pricing instead, and ChatGPT credit multipliers don’t apply.
On QCode
Calling gpt-6.1-sol on QCode is billed at the unit prices listed on /models; the Ultrafast prices on this page are OpenAI's official API prices and do not mean QCode offers that tier. Billed per token; per-model rates are listed on /models. Setup follows docs.qcode.cc: for Codex CLI set base_url = https://api.qcode.cc/openai, for the OpenAI SDK set base_url= to https://api.qcode.cc/openai/v1, and put gpt-6.1-sol in the model field. One key covers Claude, GPT and the main China-based model families, billed per token; /models lists what you can call.
Frequently asked questions
Paying and getting started from mainland China?
Alipay and WeChat Pay both work for this. QCode is prepaid: you top up first and spending is drawn from that balance. The top-up page lists Alipay (the first choice on the CNY tab), WeChat Pay, credit / debit cards (Visa / Mastercard / AMEX), Apple Pay / Google Pay, and cryptocurrency across many coins and networks. Copy a key in the console, set base_url for Codex or the Responses API to https://api.qcode.cc/openai or the OpenAI SDK base_url= to https://api.qcode.cc/openai/v1, put gpt-6.1-sol in the model field, and go by /models for unit prices, since billing is per token.
What does Ultrafast cost now, and how far is it from Standard?
The official Ultrafast row for gpt-6.1-sol is gpt-6.1-sol $12.00 $0.60 $15.00 $60.00 $24.00 $1.20 $30.00 $90.00, per million tokens: short context $12.00 input, $0.60 cached input, $15.00 cache write and $60.00 output; long context $24.00, $1.20, $30.00 and $90.00. The Standard row in the same table is gpt-6.1-sol $2.00 $0.10 $2.50 $10.00 $4.00 $0.20 $5.00 $15.00, so every one of the four unit prices is 6 times the Standard one when converted from that list. The context split follows Short context: ≤272K input tokens. Long context: >272K input tokens.
How do the 8x and the price list fit together?
They are two different meters, so do not merge them. The price list is what the API charges per token: $12.00 against $2.00 and $60.00 against $10.00, 6 times by conversion of the official list. The 8x is subscription usage: Ultrafast uses included subscription limits at 8x the Standard rate. while purchased credits and Enterprise pay-as-you-go usage are billed at 6x the Standard rate. The price list, the model page and the Codex docs checked on 2026-10-13 give no reason for the 6 and the 8 differing, and this page leaves them side by side rather than reconciling them. Running Codex with an API key follows the token prices instead: With an API key, Codex uses API token pricing instead, and ChatGPT credit multipliers don’t apply.
Which plans can use Ultrafast?
In Codex and ChatGPT Work the official line is Ultrafast is available in Codex and ChatGPT Work on Pro $500 and eligible Enterprise and Edu plans and For Enterprise workspaces, Ultrafast is off by default. The help center is blunter: Other self-serve plans, including Plus, Pro $100, Pro $200, and Business, do not include Ultrafast at launch. Purchasing credits on one of these plans does not enable it. The API side does not gate on plans: the vendor states It is available to all API users, subject to rate limits, with global processing and US and EU data residency.
What are the Ultrafast limits, and will they eat my existing allowance?
They are counted apart: Ultrafast has separate rate limits from Standard and Fast modes. The defaults are listed under GPT-6.1 Sol has the following default Ultrafast token rate limits: and read Build 1,000,000, Launch 4,000,000 and Grow 40,000,000 TPM. The organisation tier rises automatically with total credit purchases, from Build $5 in total credit purchases $500 / month up to Grow $500 in total credit purchases $200,000 / month. Subscription usage drains faster on the other meter: Ultrafast uses included subscription limits at 8x the Standard rate.
Can I call gpt-6.1-sol through QCode, and where do I see the rate?
Yes, gpt-6.1-sol is callable through QCode. One key covers Claude, GPT and the main China-based families, with the endpoints docs.qcode.cc gives: base_url = https://api.qcode.cc/openai for Codex CLI and base_url= https://api.qcode.cc/openai/v1 for the OpenAI SDK. Billing is per token and every model's rate is listed on /models. The Ultrafast numbers on this page are OpenAI's official API prices; they do not mean QCode offers that tier.
Sources
Every external fact comes from OpenAI's own pages, checked on 2026-10-13: the API changelog (developers.openai.com/api/docs/changelog, plus its .md copy), the Ultrafast tab of the official pricing page (developers.openai.com/api/docs/pricing?latest-pricing=ultrafast, plus its .md copy), the gpt-6.1-sol model page (developers.openai.com/api/docs/models/gpt-6.1-sol), the Ultrafast mode guide (developers.openai.com/api/docs/guides/ultrafast-mode), the Codex speed and tier page (developers.openai.com/codex/agent-configuration/speed), the ChatGPT & Codex changelog (developers.openai.com/codex/changelog) and the rate-limit and usage-tier page (developers.openai.com/api/docs/guides/rate-limits); help-center article 20001516 comes from a web.archive.org copy of that page. Wording about our own service and its setup comes from docs.qcode.cc (payment methods, endpoints and API paths) and from the qcode.cc pages /gpt-6-1-sol-api-guide, /gpt-5-6-sol-ultrafast and /buy-claude-api-credits-guide, checked the same day.
Want gpt-6.1-sol billed per token? Start here
One key covers Claude, GPT and the main China-based model families, billed per token; live rates are on /models, and setup is self-serve.
Related reading
GPT-6.1 Sol API guide: pricing, specs, vs GPT-6 Sol
The same model at the Standard rates: $2 input, $0.10 cached input, $2.50 cache write, $10 output, plus the 272K whole-request rule.
GPT-5.6 Sol Ultrafast: 750 tok/s preview, not a QCode tier
The earlier Ultrafast preview page for gpt-5.6-sol, whose own conclusion is that QCode lists gpt-5.6-sol at Standard, with no Ultrafast tier.
Buying Claude API credits: Console prepay and auto-reload
A purchase page: how Claude API credits are prepaid in Console, and how auto-reload and expiry work.
Prices, multipliers and limits on this page were checked on 2026-10-13 against OpenAI's official changelog, pricing page, gpt-6.1-sol model page, Ultrafast guide and Codex docs; official pages change at any time, so the official pages govern. The how-many-times figures are our conversion of the official price list; the quotes stay in the vendor's own wording. On QCode, model availability and unit prices follow /models.