EvoLink route
No verified request identity, rate or activation date is available on this page. Register for an availability alert below.
GPT-6 Luna API
OpenAILanguage ModelAvailability watch
Access unverifiedPrice unpublishedModel ID unverifiedLaunch alerts
EvoLink access
Not verified
Official model ID
Not found in checked catalog
Pricing
Not published
Last checked
2026-09-20
Can you use the GPT-6 Luna API?
As of September 20, 2026, GPT-6 Luna is not listed in the OpenAI model catalog or pricing page we checked, and EvoLink access has not been verified. If you run extraction, classification or other repeated tasks on GPT-5.6 Luna, prepare a small acceptance set before changing your queue. A new name does not establish a lower price, higher throughput or fewer retries.
GPT-6 Luna access status
Checked September 20, 2026. Product announcements, upstream API access and EvoLink access are separate milestones.
OpenAI model catalog
No entry for this exact model was found in the catalog checked. This does not rule out an internal test.
Official pricing
No model-specific pricing row was found. Existing GPT prices are not a quote for this candidate.
We will update the integration status when model-specific documentation and EvoLink request and billing verification are available.
Get the GPT-6 Luna API availability alert
Get notified when EvoLink access is verified. An upstream announcement alone does not activate this route.
Register your email
EvoLink product alerts. No access, capacity or release date is guaranteed.
Follow product updates EvoLink Discord
Know what you can actually call
The launch update should include the verified route, current rate and integration documentation so you can evaluate the model with your own tasks.
Prepare to evaluate GPT-6 Luna
These are preparation steps, not confirmed capabilities or benchmark results for the candidate.
Build a field-level acceptance set
Sample your own short and long documents, missing fields, ambiguous labels and invalid inputs. Keep expected outputs and validation rules so a schema-valid response is not mistaken for a correct result.
Measure completed records
Record accepted records per minute at your normal concurrency, queue age and P95 completion time. Keep rate-limit errors, backoff and retries in the measurement.
Budget the whole pipeline
Count failed calls, reprocessing and any escalation to another model. Compare total spend divided by accepted records, using the actual billed token categories and tool charges.
Protect queue recovery
Keep job IDs, validation failures and model versions with each result. Test a small candidate queue first and define a replay policy that cannot duplicate external writes.
GPT-6 Luna pricing and request contract
| Field | Current status | Before switching |
|---|---|---|
| Exact request ID | Not verified | Documented ID and returned model identity |
| EvoLink pricing | Not verified | Token categories, cache, tools and billing record |
| Endpoint and controls | Not verified | Required tools, streaming, schema and reasoning settings |
| Limits and reliability | Not verified | Context, output, concurrency, retry and fallback behavior |
While waiting for GPT-6 Luna: three model options
Split your queue by the quality each record requires. Keep routine records on your measured Luna baseline, evaluate Terra for cases that miss validation, and reserve a Sol trial for difficult exceptions. Current access and rates belong on the linked product page; evaluate the full processing cost before changing the routing policy.
GPT-5.6 Luna · Keep routine processing measurable
Use your existing Luna workflow as the baseline for field extraction, ticket classification and data normalization. Track semantic accuracy as well as schema validity, accepted records per minute and retries. Keep prompts, validation rules and inputs fixed for the future GPT-6 Luna comparison.
View GPT-5.6 Luna and tier pricingGPT-5.6 Terra · Evaluate a middle option
Terra is positioned to balance capability and cost. Test it on ambiguous labels, conflicting fields and other records that fail your Luna validation. Compare error recovery, queue delay and total cost before deciding whether these records should move to a separate route.
View GPT-5.6 Terra and tier pricingGPT-5.6 Sol · Trial difficult exceptions
Evaluate Sol on records requiring several reasoning steps, reconciliation across documents or tool-assisted investigation. Define an escalation budget and keep human review for unresolved cases. Check whether the additional model call reduces correction effort enough to justify its cost.
View GPT-5.6 Sol and tier pricingGPT-6 Luna: next steps
GPT-6 Luna release tracker
Check dated channel status and what counts as a confirmed release.
GPT-6 Luna vs GPT-5.6 Luna
Prepare your acceptance set and upgrade decision. No new-model benchmark is claimed.
GPT-6 Sol
The other candidate has its own access status. One route being ready does not activate the other.
Compare the GPT family
Choose among documented model options before changing production traffic.
GPT-6 Luna API FAQ
Is GPT-6 Luna the same API model as GPT-5.6 Luna?
No interchangeable identity has been verified. GPT-5.6 Luna has its own documented model ID; the new name must not silently replace that route. Track the new model separately until its identity is confirmed.
What is the GPT-6 Luna price?
No verified GPT-6 Luna rate is available in the sources checked on September 20, 2026. Do not reuse GPT-5.6 Luna pricing or assume the next generation will cost less.
Can I send a batch to GPT-6 Luna now?
This page has no verified EvoLink route or batch contract to provide. Keep your current processing path, and check the documented endpoint, concurrency and rate limits once a candidate route is verified.
Will GPT-6 Luna return the same JSON schema?
That has not been verified. Replay representative records and test both schema validity and field accuracy. Missing fields, ambiguous labels and refusal handling need explicit acceptance rules.
Does the alert guarantee capacity for my queue?
No. It is an EvoLink availability notification, not a capacity reservation or an OpenAI access program. Production volume requires a separate check of actual route limits.
What should trigger a migration from GPT-5.6 Luna?
Require acceptable field accuracy, completion latency and cost per accepted record on your own traffic sample. Retain a rollback path when errors or escalation erase an apparent saving. The linked upgrade guide explains how to set those gates.