How does GPT-5.4 rank among AI models?
GPT-5.4 is a superseded version and is not on the current board as of 22 September 2026. Its archive position, where one exists, is shown on the page; the newer version carries the current rank.
OpenAI·openai/gpt-5.4
GPT-5.4 is ModelCap archive rank #25 (Measured), from the ModelCap Index as of 22 Sept 2026, 02:49 UTC.
GPT-5.4 listed OpenRouter input price is $2.50 per 1M tokens and listed output price is $15.00 per 1M tokens, from OpenRouter listed prices as of 22 Sept 2026, 02:49 UTC.
GPT-5.4 has a published context limit of 1M tokens in the ModelCap catalogue as of 22 Sept 2026, 02:49 UTC.
GPT-5.4 has 7 published OpenRouter serving endpoints, from OpenRouter as of 22 Sept 2026, 02:49 UTC.
GPT-5.4 weight access is api only, from ModelCap classification of published repository metadata as of 22 Sept 2026, 02:49 UTC.
Snapshot facts · download the public dataset · methodology
ModelCap Index · All-versions archive
79.6#25
Measured · 70% support · score interval 75.2–84.0
Evidence as of 21 Sept 2026, 17:21 UTC
Superseded by GPT-5.5. It no longer holds a current-board position; its all-versions archive rank is #25.
Input
$2.50
per 1M tokens
Output
$15.00
per 1M tokens
Context
1M
tokens
Providers
7
endpoints
Gateway spend
—
latest · Vercel share
Providers
Uptime measured over the last 30 minutesReasoning efforts
Best supported result can inform ModelCapPublished Max, X-High, High and other configurations for this model. These are sourced scores, not head-to-head essays.
Related community configurations
Human preference ratingOther versions (6)
ModelCap Index · All-versions archive
Methodology79.6
Measured · 6 public benchmark observations across 6 boards
All-versions archive
#25
- Public rank basis
- Measured (measured)
- Index support
- 70.3% · strong
- Index interval
- 75.2–84.0
- Rank posterior
- Not available
- Top 5 / Top 10 probability
- Not available / Not available
- Posterior as of
- Snapshot timestamp unavailable
- Evidence as of
- 21 Sept 2026, 17:21 UTC
- Identity binding
- Exact catalogue product · openai/openai/gpt-5.4 · aggregates disclosed benchmark configurations · endpoint configuration not claimed
- Index observations used
- 6
- Qualified Index evidence point
- 79.6
Bayesian Evidence Score · measured evidence input only
BES normalizes admitted public benchmark evidence for measured Index rows. Its legacy score and observables prior do not define the public language rank.
- Observed capability
- 79.6
- Evidence status
- confirmed · 91% mass
- General preference
- 81.9
- Coding
- 80.0
- Agents & tools
- 50.6
- Reasoning
- 53.8
- Evidence breadth
- 100%
- Evidence coverage
- 70%
Market and catalogue signals · secondary, never the language Index rank
- Usage (OpenRouter popularity)
- 22.7
- Liquidity (providers × uptime)
- 99.5
- Open reach (HF downloads)
- —
- Surface (context / tools / modalities)
- 100.0
- Freshness
- 46.6
- · OpenRouter popularity #87
- · 3 live providers
- · Catalogue: context 1050000, reasoning, tools, modalities Text/Image/File
- · First seen 2026-03-05
These adoption and deployment observations remain context only. They do not change this model's ModelCap Index score or rank.
BES evidence selected · measured input only
- Arena Overall · High81.9
- Arena Coding · High80.0
- LMArena Agent · High47.1
- WildClawBench OpenClaw57.7
- ARC-AGI-2 Verified · X-High63.8
- ARC-AGI-3 Verified · High30.5
The bounded 0–100 capability score blends each source’s competitive placement with its published achievement, then combines capability families. Evidence breadth adds a modest uncertainty adjustment; price and popularity are not part of benchmark-led rank.
Deployment readiness
Metadata index53.8
partial · 56% metadata coverage
- Artifact reproducibility
- 0.0
- Access & legal clarity
- 30.3
- Deployability
- 99.0
- Evaluation provenance
- 90.0
Missing: artifactReproducibility.revision-sha, artifactReproducibility.architecture-and-model-type, artifactReproducibility.parameter-count, artifactReproducibility.serialized-weight-format, artifactReproducibility.base-lineage, accessLegalClarity.declared-license, accessLegalClarity.resolvable-license-terms, evaluationProvenance.evaluation-dataset-revisions.
A metadata completeness and deployability index—not a safety certification, quality grade, or production-readiness claim.
Market activity
Venue coverage- Market Gravity
- 47.2
- OpenRouter weekly popularity
- #87 of 322
Market Gravity uses OpenRouter’s full-catalogue popularity order. Sparse Vercel top-ten observations appear only when published and do not affect the score.
Arena preference
- Status
- Ranked
- Source
- LMArena
- Official rank
- #46 of 402
- Rating
- 1465.7
- 95% confidence interval
- 1462–1469
- Votes
- 64K
- Source published at
- 13 Sept 2026, 00:00 UTC
- Category
- overall
- Identity match confidence
- 100%
Market Gravity breakdown
- Usage55%
- 22.7
- Liquidity25%
- 99.5
- Open reach15%
- 50.0
- Freshness5%
- 46.6
OpenRouter weekly popularity across the full model catalogue.
Independent providers versus the model's open or closed cohort, weighted by uptime.
Hugging Face 30-day downloads for open models; neutral for closed models.
Time since first public availability, on a six-month half-life.
Specification
- Context window
- 1M tokens
- Max output
- 128K tokens
- Inputs
- Text, Image, File
- Outputs
- Text
- Tokenizer
- GPT
- Cached input tokens
- $0.25 / 1M
- First seen on OpenRouter
- 5 Mar 2026
Capabilities
- Supported: Reasoning
- Supported: Tool use
- Supported: Structured output
- Supported: Response format
- Supported: Moderated
Benchmark leaderboards featuring GPT-5.4
All leaderboardsGPT-5.4: common questions
How much does GPT-5.4 cost per 1M tokens?
GPT-5.4 is listed at $2.50 per 1M input tokens and $15.00 per 1M output tokens as of 22 September 2026; the lowest output price among 3 providers is $7.50 on OpenAI. Prices come from the OpenRouter catalogue and re-observe on every refresh. For 1,000 requests using 2,000 input and 500 output tokens each, the listed rates imply $12.50 total (2M input + 0.5M output). This excludes caching, batch discounts, prompt-length tiers, tool charges and retries; verify the chosen endpoint's rate and limits before budgeting.
What is the context window of GPT-5.4?
GPT-5.4 has a published context window of 1M tokens (1,050,000) and a maximum output of 128K tokens. Individual providers can serve less than the published maximum; the providers table lists each endpoint's own limit.
Is GPT-5.4 open-weight?
API only. No downloadable model weights are linked in the current source data; access is through a hosted API or product.
Which API providers serve GPT-5.4?
3 providers serve GPT-5.4 through OpenRouter as of 22 September 2026: OpenAI, Azure and Amazon Bedrock. Each provider's price, context limit, quantization and measured uptime are in the providers table above.
Which benchmarks has GPT-5.4 been evaluated on?
GPT-5.4 has published results on 7 public boards tracked by ModelCap as of 22 September 2026: Arena coding 1520 (#29 of 397), LMArena Agent 1.3% (#24 of 46), ARC-AGI-2 74.0% (#33 of 214), ARC-AGI-3 0.2% (#30 of 39) and AA τ²-Bench Telecom 87.1% (#84 of 436) and more. Each board page ranks every tracked model on that benchmark; the ModelCap Index combines them with published uncertainty.
When was GPT-5.4 released?
GPT-5.4 first appeared in the catalogue on 5 Mar 2026. It has since been superseded by GPT-5.5. Rank and price movements since then are recorded on the site's changes feed.