fix gpt-5.4 pricing

[mubashir1osmani]

Relevant issues

gpt-5.4 & gpt-5.4-2026-03-05

• Fixed output_cost_per_token_priority: 2.25e-05 → 3e-05 ($22.50 → $30.00/1M) — was incorrect per OpenAI docs
• Removed input_cost_per_token_above_272k_tokens_priority,
  cache_read_input_token_cost_above_272k_tokens_priority, output_cost_per_token_above_272k_tokens_priority —
  OpenAI does not publish >272K pricing for the priority tier

gpt-5.4-mini

• Fixed cache_read_input_token_cost_flex: 1e-08 → 3.75e-08 ($0.01 → $0.0375/1M)
• Fixed cache_read_input_token_cost_batches: 3.8e-08 → 3.75e-08 (rounding error)
• Added missing priority tier pricing: input 1.5e-06 ($1.50), cached 1.5e-07 ($0.15), output 9e-06 ($9.00/1M)

gpt-5.4-pro & gpt-5.4-pro-2026-03-05

• Removed all priority tier fields — gpt-5.4-pro is not listed in OpenAI's priority pricing table

ref: https://developers.openai.com/api/docs/pricing?latest-pricing=priority

cc @Sameerlite please take over this PR

Pre-Submission checklist

Please complete all items before asking a LiteLLM maintainer to review your PR

• I have Added testing in the tests/test_litellm/ directory, Adding at least 1 test is a hard requirement - see details
• My PR passes all unit tests on make test-unit
• My PR's scope is as isolated as possible, it only solves 1 specific problem
• I have requested a Greptile review by commenting @greptileai and received a Confidence Score of at least 4/5 before requesting a maintainer review

Delays in PR merge?

If you're seeing a delay in your PR being merged, ping the LiteLLM Team on Slack (#pr-review).

CI (LiteLLM team)

CI status guideline:

• 50-55 passing tests: main is stable with minor issues.
• 45-49 passing tests: acceptable but needs attention
• <= 40 passing tests: unstable; be careful with your merges and assess the risk.

• Branch creation CI run
  Link:

• CI run for the last commit
  Link:

• Merge / cherry-pick CI run
  Links:

Type

🆕 New Feature
🐛 Bug Fix
🧹 Refactoring
📖 Documentation
🚄 Infrastructure
✅ Test

Changes

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
litellm	Ready	Preview, Comment	Mar 28, 2026 10:32pm

[codspeed-hq]

Merging this PR will not alter performance

✅ 16 untouched benchmarks

---

_{Comparing mubashir1osmani:fix-gpt-5.4 (23c7e99) with main (40d4e79)}

[greptile-apps]

Greptile Summary

This PR corrects pricing data for several gpt-5.4 model variants in model_prices_and_context_window.json, aligning the values with the official OpenAI priority pricing table at developers.openai.com/api/docs/pricing.

Key changes verified against OpenAI's priority pricing table:

• gpt-5.4 / gpt-5.4-2026-03-05: output_cost_per_token_priority corrected from 2.25e-05 ($22.50/1M) to 3e-05 ($30.00/1M) ✅; long-context priority fields (above-272k) removed since OpenAI publishes no long-context pricing for the priority tier (shown as - in the table) ✅
• gpt-5.4-mini: cache_read_input_token_cost_flex corrected from 1e-08 to 3.75e-08 (50% of standard 7.5e-08, consistent with all other flex-tier ratios) ✅; cache_read_input_token_cost_batches rounding fix 3.8e-08 → 3.75e-08 ✅; missing priority tier fields added (input $1.50, cached $0.15, output $9.00 per 1M) ✅
• gpt-5.4-pro / gpt-5.4-pro-2026-03-05: All priority tier fields removed — gpt-5.4-pro does not appear in OpenAI's published priority pricing table ✅

No tests were added, and the pre-submission checklist item "Adding at least 1 test is a hard requirement" is unchecked. For a data-only JSON pricing fix this is a minor process gap, but the correctness of the values is well-supported by the referenced official pricing page.

Confidence Score: 5/5

Safe to merge — all changed pricing values are verified correct against the official OpenAI priority pricing table.

All numeric corrections are cross-checked against the official OpenAI pricing docs: gpt-5.4 priority output $30/1M, gpt-5.4-mini priority tier fully added at correct rates, gpt-5.4-pro priority tier removal is justified. The only gap is a missing test, which is a process concern and does not affect correctness for a data-only JSON change.

No files require special attention — this is a single-file data fix with verified correct values.

Important Files Changed

Filename	Overview
model_prices_and_context_window.json	Pricing corrections for gpt-5.4, gpt-5.4-2026-03-05, gpt-5.4-mini, gpt-5.4-pro, and gpt-5.4-pro-2026-03-05 — all changed values are verified correct against the official OpenAI pricing docs.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[API Request with service_tier] --> B{Tier?}
    B -- standard --> C[Standard pricing<br/>input / cached / output]
    B -- flex --> D[Flex pricing ~50%<br/>input_flex / cached_flex / output_flex]
    B -- batch --> E[Batch pricing ~50%<br/>input_batches / cached_batches / output_batches]
    B -- priority --> F{Model supports<br/>priority?}
    F -- Yes --> G[Priority pricing 2x<br/>input_priority / cached_priority / output_priority]
    F -- No --> H[No priority pricing<br/>fields defined]

    subgraph "After this PR"
        G1["gpt-5.4: output_priority $30.00 was $22.50"]
        G2["gpt-5.4-mini: full priority tier added"]
        H1["gpt-5.4-pro: priority tier removed - not in OpenAI table"]
    end

    G --> G1
    G --> G2
    H --> H1

Loading

_{Reviews (1): Last reviewed commit: "fix gpt-5.4 pricing" | Re-trigger Greptile}

[Sameerlite]

.

[Classic298]

@Sameerlite we need this

[mubashir1osmani]

@ishaan-berri @ishaan-jaff