OpenAI Announced Sep 3, 2026

GPT-6 Astra

OpenAI's next flagship targets complex coding, research, and professional workflows. Compare its published results with independent measurements and check MiniRouter's current API pricing.

Context window
1,050,000 tokens
OpenAI documents up to 922,000 input tokens and 128,000 output tokens.
Reasoning
low through max
low, medium, high, xhigh, max. The none and minimal settings are unsupported.
Tool calling
Responses required
OpenAI supports text Chat Completions; Astra tool calls require Responses. Gateway and client compatibility must be checked separately.
Sampling
Remove temperature / top_p
Also remove top_logprobs; Chat Completions does not accept logprobs.

Publisher pricing

Pricing below comes from the linked publisher sources and does not represent a MiniRouter route or price.

Publisher-listed API price phases in US dollars per million tokens.
PhaseInput / 1MOutput / 1MBlended / 1MEffectiveSource
OpenAI Standard list price$10.00$50.00$20.00Sep 3, 2026open-endedOpenAI

Publisher-reported benchmarks

Results are grouped by unit. Percentage, Elo, and task-count results never share a scale.

BenchmarkReported resultReported byAs ofSource
Terminal-Bench 4.057.9%OpenAISep 3, 2026Announcement ↗
AutomationBench41.4%OpenAISep 3, 2026Announcement ↗
GPQA Diamond96.0%OpenAISep 3, 2026Announcement ↗
  • The announcement is dated September 3; MiniRouter's catalog feed dates this arrival September 4. Catalog availability is refreshed separately from this article.
  • The price table shows OpenAI's Standard list rates. Use the linked model page for current MiniRouter customer rates. OpenAI applies higher full-request rates above 272K input tokens; cache writes and Fast mode have separate pricing.
  • OpenAI documents async tools and steering for Astra. Their availability in OpenAI's API does not establish MiniRouter support for those features.
  • Artificial Analysis labels one evaluation non-reasoning. That source label is not a supported Astra API effort setting; OpenAI documents low through max.

Independent measurements

Artificial Analysis runs its own evaluations and timing against each model's first-party API. These are the only numbers on this page not reported by the publisher.

GPT-6 Astra (max) ranks #2 of 451 model families on their Intelligence Index. Their headline uses the highest-scoring effort setting; every listed setting is below.

Intelligence Index
54.7
Composite of their evaluations
Coding Index
76.9
Coding evaluations only
Blended price
$20
USD per 1M tokens, 3:1 input to output, first-party list
Output speed
Not measured
Median tokens per second
First answer token
Not measured
Median seconds, including reasoning
SettingIntelligence IndexCoding IndexBlended priceOutput speedFirst answer token
GPT-6 Astra (max)54.776.9$20Not measuredNot measured
GPT-6 Astra (xhigh)54.375.9$20Not measuredNot measured
GPT-6 Astra (high)53.477.1$20Not measuredNot measured
GPT-6 Astra (medium)52.276.7$20Not measuredNot measured
GPT-6 Astra (low)49.375.7$20Not measuredNot measured
GPT-6 Astra (non-reasoning)47.876.2$20Not measuredNot measured

Source: Artificial Analysis, read Sep 4, 2026. Measured on the model's first-party API, not a MiniRouter route.

What is not independently confirmed

  • Publisher-reported results and independent measurements are labeled separately. MiniRouter did not run these benchmarks.
  • Benchmark harnesses are not standardised. Results sharing a name across vendors have not been confirmed to use identical runs.
  • Publisher pricing can change, and introductory phases have stated end dates.
  • GPT-6 Astra is routable through MiniRouter today; live pricing on its model page is authoritative over the publisher figures here.

Related coverage

Other release dossiers and the current MiniRouter catalog.

Sources

  1. OpenAI GPT-6 Astra announcement

    Published Sep 3, 2026

    Read at OpenAI
  2. OpenAI GPT-6 Astra model documentation

    Published Sep 4, 2026

    Read at OpenAI
  3. Artificial Analysis model page

    Published Sep 4, 2026

    Read at Artificial Analysis

Editorial reference revised 2026-09-10. Catalog status comes from the generated routing snapshot.