Current State: gated agents, repository-managed Claude resources, and image API migration work
The week starting 2026-09-07 is about limited agent availability, operational controls, and migration checks rather than broad new defaults.
Key takeaways
- OpenAI GPT-6 Astra was introduced on 2026-09-03 for a limited set of organizations and was not yet generally available.
- Anthropic Claude Fable 5.1 launched on 2026-09-01 with a 1M token context window, 128k max output tokens, and $10 / $50 USD per MTok pricing.
- xAI said on 2026-09-02 that `grok-imagine-image-quality` will be retired on 2026-11-02 and served by `grok-imagine-image-2.0` with `quality` set to `low`.
- OpenAI Healthcare Public Data in ChatGPT for Clinicians became available on 2026-09-01 for eligible United States users and uses nine read-only apps that do not access patient charts.
- Anthropic ant CLI 1.30.0 added `ant apply` on 2026-09-03 to manage agents, environments, skills, memory stores, and deployments from repository files.
Review
OpenAI, Anthropic, and xAI all published changes that affect how operations teams should govern AI systems in production workflows. OpenAI's GPT-6 Astra is still limited and not generally available, Anthropic's Claude Fable 5.1 has specific tool-use and thinking-block constraints, and xAI has a dated retirement path for an image model slug.
Dated events and sources
OpenAI: Introducing GPT-6 Astra
Fact: On 2026-09-03, OpenAI introduced GPT-6 Astra for coding, research, computer use, and complex multi-step work, with access rolling out to a limited set of organizations and Astra not yet generally available.
Nova3 operational read: Treat GPT-6 Astra as a gated evaluation path, not a production baseline. If your organization gets access, test document, spreadsheet, presentation, and computer-use workflows with review steps because OpenAI says conversations may be paused or stopped when safety monitoring detects a potential instruction-interpretation issue.
Uncertainty: Pricing, the exact list of eligible organizations, region limits, and the general availability date are not established.
OpenAI ChatGPT release notes: Introducing GPT-6 AstraVendor source headline: Introducing GPT-6 Astra
OpenAI: Zendesk & OneNote Plugins in ChatGPT and Codex [Beta]
Fact: On 2026-09-03, OpenAI added the Zendesk and OneNote plugins in the Plugin directory for use in ChatGPT and Codex, with Zendesk using members' own Zendesk accounts and OneNote supporting note search, summaries, decisions, action items, and supported create or update actions.
Nova3 operational read: Support teams can evaluate Zendesk workflows without granting broader access than each member already has in Zendesk. Knowledge-management teams should review which OneNote actions are supported before allowing create or update use.
Uncertainty: Pricing, region limits, tenant eligibility, and whether Beta changes production support terms are not established.
OpenAI ChatGPT release notes: Zendesk & OneNote Plugins in ChatGPT and Codex [Beta]Vendor source headline: Zendesk & OneNote Plugins in ChatGPT and Codex [Beta]
Model Context Protocol and connected tools · Evaluate an AI implementation partner
OpenAI: Healthcare Public Data in ChatGPT for Clinicians
Fact: On 2026-09-01, OpenAI made Healthcare Public Data in ChatGPT available to eligible ChatGPT for Clinicians users in the United States, with nine read-only apps for public healthcare sources that do not access patient charts.
Nova3 operational read: Clinician operations teams should treat the plugin as a public-source lookup tool, not a patient-record tool. Add usage guidance that repeats OpenAI's instruction not to include protected health information in searches sent to public sources.
Uncertainty: Pricing, availability outside the United States, exact eligibility criteria, and whether the feature is preview are not established.
OpenAI ChatGPT release notes: Healthcare Public Data in ChatGPT for CliniciansVendor source headline: Healthcare Public Data in ChatGPT for Clinicians
Anthropic: Version 1.30.0 of the ant CLI adds ant apply
Fact: On 2026-09-03, Anthropic released version 1.30.0 of the `ant` CLI with `ant apply`, which creates and updates agents, environments, skills, memory stores, and deployments from files in a repository and writes a `claude-lock.json` lockfile.
Nova3 operational read: Platform teams should evaluate `ant apply` for change-controlled Claude resource management through repository files and CI. Commit the generated lockfile if you want later runs to update the same resources instead of creating new ones.
Uncertainty: Pricing, region limits, account eligibility, and whether `ant apply` is preview are not established.
Anthropic release notes: Version 1.30.0 of the ant CLI adds ant applyVendor source headline: Version 1.30.0 of the ant CLI adds ant apply
Anthropic: Claude Fable 5.1 and Claude Mythos 5.1
Fact: On 2026-09-01, Anthropic launched Claude Fable 5.1 (`claude-fable-5-1`) and Claude Mythos 5.1 (`claude-mythos-5-1`) with a 1M token context window by default, 128k max output tokens, always-on adaptive thinking, $10 / $50 USD per MTok pricing, and $0.25 per MTok cache reads.
Nova3 operational read: Teams using long-context agentic coding or research workflows should benchmark Claude Fable 5.1 against current prompts and cache patterns before migration. Cost models should separate base input, output, and cache-read volume because Anthropic changed cache-read pricing for these models.
Uncertainty: General availability and region limits for Claude Mythos 5.1 outside Project Glasswing participants are not established.
Anthropic release notes: Claude Fable 5.1Vendor source headline: Claude Fable 5.1 and Claude Mythos 5.1
AI operations, explained · Nova3 implementation and advisory rates
Anthropic: tool_choice behavior on Claude Fable 5.1 and Claude Mythos 5.1
Fact: On 2026-09-01, Anthropic said `tool_choice` types `any` and `tool` are not supported on Claude Fable 5.1 and Claude Mythos 5.1 and return a 400 error, while `auto` and `none` are unchanged.
Nova3 operational read: Audit tool-calling requests before moving traffic to Claude Fable 5.1 or Claude Mythos 5.1. Use strict tool use or structured outputs when schema-conformant tool inputs are required.
Uncertainty: Pricing impact, rollout timing by platform, and whether Anthropic will later support `any` or `tool` are not established.
Anthropic release notes: tool_choice on Claude Fable 5.1 and Claude Mythos 5.1Vendor source headline: tool_choice behavior on Claude Fable 5.1 and Claude Mythos 5.1
Anthropic: Thinking blocks on Claude Fable 5.1 and Claude Mythos 5.1
Fact: On 2026-09-01, Anthropic said thinking blocks produced by Claude Fable 5.1 and Claude Mythos 5.1 are preserved only for the model that produced them or a newer one, and new accounts created on or after 2026-08-31 receive a 400 error on Claude Fable 5.1 if a replayed thinking block follows changes to the `system` prompt, `tools`, or an earlier message.
Nova3 operational read: Do not assume thinking-block replay is portable across older Claude models. Regression tests should cover prompt, tool, and earlier-message changes for accounts subject to the 2026-08-31 rule.
Uncertainty: Pricing impact, platform-specific rollout timing, and long-term support behavior for the `thinking-binding-controls-2026-08-01` beta header are not established.
Anthropic release notes: Thinking blocks on Claude Fable 5.1 and Claude Mythos 5.1Vendor source headline: Thinking blocks on Claude Fable 5.1 and Claude Mythos 5.1
Model Context Protocol and connected tools · Evaluate an AI implementation partner
xAI: grok-imagine-image-quality retirement on November 2
Fact: On 2026-09-02, xAI said `grok-imagine-image-quality` is retired on 2026-11-02, and requests to the slug will be served by `grok-imagine-image-2.0` with `quality` set to `low`, no change to the request or response shape, and a lower per-image price.
Nova3 operational read: Inventory any calls to `grok-imagine-image-quality` and test output expectations against `grok-imagine-image-2.0` with `quality` set to `low`. Update internal runbooks before 2026-11-02 even though xAI says the request and response shape will not change.
Uncertainty: Exact per-image prices, region limits, and whether any temporary compatibility period exists after 2026-11-02 are not established.
xAI release notes: grok-imagine-image-quality retirement on November 2Vendor source headline: grok-imagine-image-quality retirement on November 2
AI operations and integration maintenance · Discuss an integration or migration
xAI: Grok 4.6
Fact: On 2026-08-28, xAI made Grok 4.6 available on the xAI API with a 500k context window, text and image inputs, text-only output, no text output limit, and pricing of $2 / $0.50 / $6 per 1M tokens below 200k prompt tokens and $4 / $1 / $12 above 200k prompt tokens.
Nova3 operational read: Teams considering Grok 4.6 should model costs around the 200k prompt-token pricing threshold. Image-input workflows should confirm that text-only output fits the application before migration.
Uncertainty: Region limits, account eligibility, rate limits, and whether Grok 4.6 is preview are not established.
xAI release notes: Grok 4.6Vendor source headline: Grok 4.6
AI operations, explained · Nova3 implementation and advisory rates
Questions this edition answers
Should an operations team plan production workflows around OpenAI GPT-6 Astra now?
No, OpenAI GPT-6 Astra should not be treated as a generally available production dependency. OpenAI said on 2026-09-03 that access is rolling out to a limited set of organizations and that Astra is not yet generally available. Teams with access should validate templates, document generation, computer-use steps, and review behavior before relying on it.
What should Claude Fable 5.1 buyers check before migrating?
Claude Fable 5.1 buyers should check token costs, cache-read economics, output limits, tool-choice behavior, and thinking-block replay behavior. Anthropic said on 2026-09-01 that Claude Fable 5.1 supports a 1M token context window, 128k max output tokens, and $10 / $50 USD per MTok pricing. Anthropic also said `tool_choice` types `any` and `tool` return a 400 error on Claude Fable 5.1 and Claude Mythos 5.1.
Does OpenAI Healthcare Public Data in ChatGPT for Clinicians access patient charts?
No, OpenAI says the Healthcare Public Data apps are read-only and do not access patient charts. OpenAI said on 2026-09-01 that eligible ChatGPT for Clinicians users in the United States can use nine apps for public healthcare sources. OpenAI also says users should not include protected health information in searches sent to public sources.
What is the xAI `grok-imagine-image-quality` migration deadline?
The xAI `grok-imagine-image-quality` slug is retired on 2026-11-02. xAI said on 2026-09-02 that requests to the slug will be served by `grok-imagine-image-2.0` with `quality` set to `low`, with no change to the request or response shape. Operations teams should test image output expectations and cost assumptions before the retirement date.
Editorial method
Current State is a weekly review. Nothing appears here without a first-party source and an explicit event date.
/current-state.md mirrors the latest edition.