April 23, 2026. OpenAI released GPT-5.5 — six weeks after GPT-5.4, continuing the company's aggressive release cadence as frontier AI labs compete for enterprise customers.
The internal codename is "Spud." The public positioning is "a new class of intelligence for real work."
GPT-5.5 is available today for Plus, Pro, Business, and Enterprise subscribers in ChatGPT and Codex. API access is coming "very soon" but requires different safeguards that aren't ready at launch.
What Changed from GPT-5.4
OpenAI describes GPT-5.5 as a "faster, sharper thinker for fewer tokens." The core improvement isn't a dramatic capability jump — it's efficiency. The model achieves better results than GPT-5.4 while using fewer tokens to get there.
Greg Brockman, OpenAI co-founder, described it as "a big step towards more agentic and intuitive computing" — the model understands intent faster and can carry more of the work itself with less user hand-holding.
Mark Chen, VP of Research at OpenAI, noted it "shows meaningful gains on scientific and technical research workflows" — specifically drug discovery and early-stage scientific work.
The combination: same intelligence tier as GPT-5.4, improved token efficiency, and better handling of multi-step autonomous tasks.
Key Capabilities
Agentic Task Handling
GPT-5.5 is primarily positioned for multi-step workflows — tasks that require planning, tool use, and self-checking across multiple iterations. It outperforms GPT-5.4 at:
- Agentic coding — writing, debugging, and verifying code across long sessions
- Computer use — operating software interfaces autonomously
- Knowledge work — research synthesis, document analysis, report generation
- Scientific research — particularly early-stage drug discovery and technical workflows
The improvement isn't a single capability — it's reduced friction across the full workflow: fewer clarifying questions needed, better task decomposition, more reliable self-correction.
Codex: 400K Context Window
For subscribers using Codex (OpenAI's AI coding environment), GPT-5.5 ships with a 400K token context window — large enough to hold entire codebases in context.
Codex also gains a Fast mode: 2.5x the cost, 1.5x faster token generation. Useful for latency-sensitive development workflows where waiting on each step becomes a bottleneck.
GPT-5.5 outperforms its predecessors and current competitors across multiple benchmarks:
- Beats GPT-5.4 across agentic coding, knowledge work, and scientific research
- Beats Gemini 3.1 Pro on the same categories
- Beats Claude Opus 4.5 on core benchmarks (notably, Anthropic shipped Claude Opus 4.7 one week earlier — GPT-5.5 will need to be tested directly against it)
OpenAI hasn't published a full benchmark table at launch — unlike Anthropic's detailed comparisons at the Opus 4.7 release last week.
Availability by Tier
| Plan | GPT-5.5 | GPT-5.5 Pro | GPT-5.5 Thinking |
|---|
| Free | — | — | — |
| Plus ($20/mo) | Yes | — | Yes |
| Pro ($200/mo) | Yes | Yes | Yes |
| Business | Yes | Yes | Yes |
| Enterprise | Yes | Yes | Yes |
GPT-5.5 Pro is the highest-capability variant, available only on Pro, Business, and Enterprise plans.
GPT-5.5 Thinking provides extended reasoning — available from Plus and above.
API access: Not available at launch. OpenAI says it's coming "very soon" but requires additional safeguards review before release.
API Pricing (When Available)
Pricing is set at 2x the cost of GPT-5.4:
| Token Type | GPT-5.5 | GPT-5.4 |
|---|
| Input | $5.00 / 1M tokens | $2.50 / 1M tokens |
| Cached Input | $0.50 / 1M tokens | $0.25 / 1M tokens |
| Output | $30.00 / 1M tokens | $15.00 / 1M tokens |
The doubling reflects both the model's capability improvements and OpenAI's positioning of GPT-5.5 as an enterprise-tier model. Developers building cost-sensitive applications may want to evaluate whether GPT-5.4 or a lighter model serves their use case better before switching.
Safety
GPT-5.5 launches with what OpenAI calls its "strongest set of safeguards to date." The process included:
- Internal red-teaming
- External red-teaming
- Feedback from nearly 200 early-access partners before release
The delayed API launch is explicitly tied to safety review. OpenAI stated that API deployments "require different safeguards" and that those aren't ready yet — a notable public acknowledgment given the usual pressure to ship.
Context: OpenAI's Release Pace
GPT-5.5 shipped six weeks after GPT-5.4. The timeline matters because it signals a fundamental change in how OpenAI is developing and releasing models — away from large infrequent jumps toward continuous incremental updates.
This creates a moving target for competitors. Anthropic shipped Claude Opus 4.7 one week ago. Google's Gemini 3.1 Pro is the current third-party benchmark baseline. All three companies are shipping on timelines measured in weeks rather than months.
OpenAI's broader strategic direction with GPT-5.5 is toward a "superapp" — a unified platform combining ChatGPT, Codex, and an AI browser into a single enterprise service. GPT-5.5 is described as one step toward that goal.
Chief Scientist Jakub Pachocki noted in internal communications that recent progress has been "surprisingly slow" — suggesting the team believes there's significantly more headroom and that the current pace is not the ceiling.
How It Compares: GPT-5.5 vs Claude Opus 4.7
| GPT-5.5 | Claude Opus 4.7 |
|---|
| Release date | April 23, 2026 | April 16, 2026 |
| Input price | $5.00/1M | $5.00/1M |
| Output price | $30.00/1M | $25.00/1M |
| Context window | Standard (Codex: 400K) | 200K |
| Vision | Standard | 3.75MP (3× upgrade) |
| Thinking / Extended reasoning | Yes | Yes (xhigh effort level) |
| API availability | Coming soon | Available now |
| Benchmarks (agentic coding) | Strong | 87.6% SWE-bench Verified |
| Document reasoning | Strong | 80.6% OfficeQA Pro |
The models are closely matched at the top tier. GPT-5.5 wins on Codex context window; Claude Opus 4.7 has a published benchmark table and cheaper output tokens. Direct head-to-head data will emerge as independent benchmarks run.
Bottom Line
GPT-5.5 is an efficiency upgrade more than a capability leap — the same intelligence tier as GPT-5.4, but faster, more token-efficient, and better tuned for autonomous multi-step work.
For ChatGPT subscribers already on Plus or above, it's available now in ChatGPT and Codex. For developers building on the API, the wait continues — but "very soon" suggests days to weeks, not months.
The 2x pricing is the main friction point. Teams currently running GPT-5.4 in production should evaluate whether the efficiency gains offset the cost increase before migrating.
ChatGPT access: chatgpt.com — Plus and above, available now.
API: Not yet available. Monitor platform.openai.com for the release announcement.
Sources:
