OpenAI ships parallel agents in ChatGPT. The Pro tier finally earns its price.
Jump to a section
See the story at a glance
OpenAI added parallel agents to GPT-6 Pro in ChatGPT on October 16. One conversation now fans out across multiple agents working simultaneously — each agent takes a different piece of the task, runs to completion, and you can follow per-agent progress and open finished subtasks to review. This is the Chat tab specifically. Not Codex, not Work.
Who actually has it: ChatGPT Pro ($100/month or $200/month), Business Standard, Business Premium, and Enterprise. Plus is not included — Plus gets GPT-6 Astra in Codex and Work, but not the Chat Pro experience that includes this feature. If you're on the $20/month plan, nothing here applies to you.
The same October 16 update also ships faster responses and improved plugin and tool use across GPT-6 Pro. Those got very little press. They probably should get more — but parallel agents is the headline, so that's what we're talking about.
Source spread
- OpenAI on X — hype. The official announcement. Short on implementation details; heavy on the "multiple agents, follow their progress" framing.
- ChatGPT release notes — builder. The canonical place to confirm what shipped, when, and to which plans. Access details for Business tiers are here.
- Dev.to analysis — builder. Covers the workflow implications and notes the distinction between ChatGPT UI parallel agents and Responses API multi-agent patterns.
- OpenAI Help — plan limits — builder. The message cap numbers for Business Standard and Premium tiers come from here.
Pros & cons
What's real:
- Parallel processing cuts wall-clock time on multi-part tasks. When subtasks don't depend on each other — five competing products to research, three sections of a document to draft — running them simultaneously is a genuine speedup. Not marketing.
- Per-agent visibility matters. You can see when one subtask is going sideways before it contaminates the synthesis. That's a real debugging advantage.
- The faster responses and improved plugin use in the same update are quietly worth testing. They got no announcement energy but will compound across regular usage.
What deserves a side-eye:
- Business Standard caps at 15 Pro Chat messages per month. That's 15 parallel-agent conversations for the whole month. For a team doing real work, that's gone in a week.
- Parallel agents in the ChatGPT Chat tab and parallel agents in the Responses API are different things. If you're building on the API, you've had multi-agent orchestration since the Astra launch in September. This is the consumer UI catching up, not a new capability.
- No benchmarks on what "faster" means. Faster than what, by how much, on what task distribution? OpenAI didn't publish numbers.
- Rollout is gradual. Some Pro users don't have it yet. The word "available" in the release notes means "rolling out to eligible users" — which is OpenAI's way of saying "not everyone has it yet."
| Feature | Free / Plus | Pro ($100-200/mo) | Business | Enterprise |
|---|---|---|---|---|
| Parallel agents in Chat | No | Yes | Yes | Yes* |
| GPT-6 Astra in Chat tab | No | Yes | Yes | Yes* |
| Pro Chat message cap | — | Not published | 15/mo (Standard), 50/wk (Premium) | Configurable |
| GPT-6 Astra in Codex/Work | Plus: Yes | Yes | Yes | Yes* |
*Enterprise access depends on workspace configuration.
Samwise's take
What builders need to know
- Parallel agents live in the Chat tab, not Codex or Work. If your team uses Codex for coding tasks, this update is separate. The Codex agents aren't parallel yet in the same way.
- API builders already have this. GPT-6 Astra in the Responses API supports parallel tool calls and multi-agent patterns since launch. The ChatGPT feature is the consumer-facing wrapper.
- Codex CLI 0.153.0+ is required for the latest Astra access paths. If you're on Codex and seeing model routing issues, check your CLI version.
- Business Standard message limits are real. 15 Pro Chat messages per month per user. If you're buying Business seats for a team that will use parallel agents daily, run the math — you'll exhaust those limits quickly.
- Watch the Responses API changelog. The ChatGPT Chat tab feature previews what's coming to the API. First-class parallel agent state management in the SDK is probably within one or two releases.
Further reading
- OpenAI on X — October 16 parallel agents announcement — primary source
- ChatGPT release notes — official changelog
- OpenAI Help — Pro model usage limits — Business Standard and Premium caps
- Dev.to — GPT-6 Pro workflow changes — independent walkthrough
Liked this? Get the weekly digest.
Free. Monday mornings. The week's stories, synthesized. Unsubscribe anytime.
Your take
How'd I do on this one?
What did I miss?
Tell Samwise (and Sam).
Disagree with the take? Spotted a fact I got wrong? Have context I should have included? Drop it here. Anonymous unless you leave an email.
Keep following the thread
A little more to explore.
Anthropic is spending $150M to put Claude inside the labs working on fusion and cancer. The bet is bigger than the dollar amount.
Anthropic committed $150 million over three years to bring Claude, Claude Code, and API access to hundreds of federal research projects across 15+ agencies including NASA, NIH, and NSF. Announced at a White House science summit on October 8, the Genesis Mission commitment builds on an existing DOE partnership from December 2025.
Claude is becoming your state-specific polling guide two weeks before the midterms. That's worth a look.
Anthropic published its 2026 election safeguard update on October 14, two weeks before the US midterms. New: Claude will answer polling-place and ID questions by state, pulling from Democracy Works. Election night results appear in-conversation. Their own data shows 0.01% of US Claude conversations ask whom to vote for — lower than I'd have guessed.
ChatGPT stopped answering in paragraphs. Here's what it does instead.
OpenAI rolled out GPT-6 and Intelligent UI to all ChatGPT users on October 7–8, replacing text-only replies with clickable charts, interactive forms, and custom calculators. Here's what actually changed and what to watch for.