Indlæser kataloget…
OpenAI · Gpt 5-modelfamilien
gpt-5.5-pro
Udgivet 23. apr. 2026
OpenAI's closed GPT-5.5 Pro reasoning model with configurable reasoning effort and a 1,050,000-token context window.
27,22€/163,34€per 1 mio. ind / ud
Indlæser kataloget…
OpenAI's closed GPT-5.5 Pro reasoning model with configurable reasoning effort and a 1,050,000-token context window.
| Inferensleverandør | Suverænitet | Kører i | Ind / ud per 1 mio. | Status | Fastgør |
|---|---|---|---|---|---|
| L0 | 27,22 / 163,34€+3 betingede priser | Online |
14 anvendelsesområder klassificeret af redaktøren. Noten forklarer hvorfor.
Pro-tier flagship reasoning variant from OpenAI; multi-step code-changing workflows are the core target for this tier. No context window or benchmark is published in the card, so this rests on tier positioning rather than measured scores.
Frontier reasoning tier suited to detecting logic and security issues across code. Treat as provisional until benchmark or context data is curated.
The Pro tier is positioned for extended, multi-step reasoning, which matches planning workloads. Confidence is limited because the card lists no capabilities or benchmark data.
High-reasoning tier fits multi-source synthesis, where intermediate reasoning quality matters more than throughput. The card has no context window or evaluation data to confirm this.
Function calling is expected for an OpenAI frontier model, but the card does not list it under capabilities. Confirm support before committing.
Reasoning-tier model suited to analytical work. Code-execution or data-tool support is not confirmed in the card, so the fit is moderate.
Frontier general-purpose model with broad writing capability. Output-length and context limits are unpublished, so the fit is moderate.
General capability is adequate for summarisation, but this Pro tier is higher-cost than needed for routine summaries. Context window is unpublished.
Strong instruction-following is expected at this tier, but structured-output support is not flagged in the card. Confirm JSON or schema support before relying on it.
Reasoning capability suits dense regulatory text, but the CLOUD Act exposure (US jurisdiction) may be a constraint for regulated data. Evaluate against the customer's sovereignty requirements.
Hosted US-jurisdiction model (CLOUD Act exposure: yes) and high-cost tier make it a poor default for routine PII handling. A smaller, deployable model is usually the better choice.
No audio or speech-generation capability is listed for this model. Prefer a dedicated speech model unless audio output is confirmed.
No audio-input capability is listed for this model. A dedicated speech-to-text model is the more reliable choice for transcription.
Tal fra eksterne kilder, ikke vores egne vurderinger. Percentiler er målt mod hele Stav-kataloget.
curl https://api.stav.ai/v1/chat/completions \
-H "Authorization: Bearer $STAV_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "openai/gpt-5.5-pro",
"messages": [{"role": "user", "content": "Hello!"}]
}'→ X-Stav-Request-Id · X-Stav-Model · X-Stav-Provider → X-Stav-Sovereignty-Level — det niveau, forespørgslen blev betjent på