All times Eastern (ET)Provider pages checked daily at
ProSign inRSS

Migrating from gpt-4.1 to gpt-4.5-preview

What changes when you move from gpt-4.1 to gpt-4.5-preview: context window 1,047,576 → 128,000 tokens.

Status
Active
Shutdown
none announced
Time left
—
Status
Retired
Shutdown
Jul 14, 2025
Time left
—

gpt-4.5-preview is itself retired. Its replacement is gpt-4.1.

Prices

USD per 1M tokens, standard tier
Typegpt-4.1gpt-4.5-previewChange
Input$2.00first seen October 1, 2026 · sourceNot published—
Cached input$0.50first seen October 1, 2026 · sourceNot published—
Output$8.00first seen October 1, 2026 · sourceNot published—

Other tiers, prompt-size bands and modalities are on each model's page.

Limits

Tokens
Limitgpt-4.1gpt-4.5-previewChange
Context window1,047,576 tokensin place when tracking began (October 6, 2026) · source128,000 tokensin place when tracking began (October 6, 2026) · source−87.8%
Max output tokens32,768 tokensin place when tracking began (October 6, 2026) · source16,384 tokensin place when tracking began (October 6, 2026) · source−50%

Benchmarks

Same benchmark, source and setting on both sides
Benchmarkgpt-4.1gpt-4.5-previewChange
Arena TextArena. Ranges don't overlap.1415 ±41445 ±6+30
Arena Text: CodingArena.1456 ±71458 ±13+2
within range
Arena VisionArena.1214 ±71225 ±12+11
within range
GPQA DiamondEpoch AI.66.9%68.7%+1.8 pts
MATH Level 5Epoch AI.83.0%78.6%−4.4 pts
OTIS Mock AIME 2024-2025Epoch AI.38.3%37.8%−0.5 pts

A score from one source is never compared with another source's, and company-reported scores are compared only within the same company; settings named where they differ. Data: Arena (arena.ai), CC BY 4.0; Epoch AI, Capabilities & benchmarking (epoch.ai/benchmarks), CC BY.

Features and parameters

Compare models
  • Fine-tuning: supported on gpt-4.1, not supported on gpt-4.5-preview.gpt-4.1: in place when tracking began (October 6, 2026) · source · gpt-4.5-preview: in place when tracking began (October 6, 2026) · source

Same on both

  • Structured outputs: supported on both.gpt-4.1: in place when tracking began (October 6, 2026) · source · gpt-4.5-preview: in place when tracking began (October 6, 2026) · source
  • Function calling: supported on both.gpt-4.1: in place when tracking began (October 6, 2026) · source · gpt-4.5-preview: in place when tracking began (October 6, 2026) · source
  • Prompt caching: supported on both.gpt-4.1: in place when tracking began (October 6, 2026) · source · gpt-4.5-preview: in place when tracking began (October 6, 2026) · source
  • Batch API: supported on both.gpt-4.1: in place when tracking began (October 6, 2026) · source · gpt-4.5-preview: in place when tracking began (October 6, 2026) · source
  • Image input: supported on both.gpt-4.1: in place when tracking began (October 6, 2026) · source · gpt-4.5-preview: in place when tracking began (October 6, 2026) · source

Not published for gpt-4.1: PDF input, Reasoning / thinking, temperature parameter, top_p parameter, logprobs.

Not published for gpt-4.5-preview: PDF input, Reasoning / thinking, Built-in web search, temperature parameter, top_p parameter, logprobs.

Aliases

All aliases

No aliases point at either model.

On cloud platforms

All platform dates
ModelPlatformPlatform IDShutdownTime left
gpt-4.1Microsoft Foundryversion 2025-04-14gpt-4.1Apr 14, 2027Deprecated181 days

Read daily from each platform's own pages. The status above uses each company's own date. Time left counts to 00:00 UTC on the shutdown date, which is the evening before in ET.

Checklist

  1. Change the model string from gpt-4.1 to gpt-4.5-preview.
  2. Check max_tokens: gpt-4.5-preview returns at most 16,384 tokens, gpt-4.1 up to 32,768 tokens.
  3. Check prompt sizes: gpt-4.5-preview's context window is 128,000 tokens, gpt-4.1's 1,047,576 tokens.

Compared from AIWire's published data when this page loaded. Each fact links to the page it came from. Model pages: gpt-4.1, gpt-4.5-preview.