GPT-5.4
Official information
- Release date
- Mar 5, 2026
- Official model ID
- gpt-5.4
- Vendor source
- GPT-5.4
- Identity verified
- Aug 13, 2026
API pricing
- Verification
- Retained last valid price
- Verified at
- Value benchmark tier
- Fixed USD rate
| Tier / region | Input / output · per MTok | Context pricing tier | Billing | Validity | Source |
|---|---|---|---|---|---|
| Fixed USD rate | $2.5 / $15 | — | PAYG | — | Official rates |
Rankings by category
Overall
#14High 0.9369.0
| Source | Percentile | Raw score | Config folding |
|---|---|---|---|
| Epoch AI | 76.5 | 157.3 | Single config |
| LiveBench | 70.7 | 78.8 | Single config |
| LMArena | 59.8 | 1460.7 | Median of 2 configs |
Cross-source percentile spread 16.7
Coding
#27High 0.7845.8
| Source | Percentile | Raw score | Config folding |
|---|---|---|---|
| DeepSWE | 37.5 | 51.8 | Single config |
| LiveBench | 62.0 | 63.3 | Single config |
| LMArenatext-coding | 57.8 | 1487.5 | Median of 2 configs |
| · webdev | 18.1 | 1443.8 | Median of 3 configs |
Cross-source percentile spread 24.5
Agent
#24High 0.8335.7
| Source | Percentile | Raw score | Config folding |
|---|---|---|---|
| Agents' Last Exam | 33.9 | 33.0 | Median of 2 configs |
| DeepSWE | 37.5 | 51.8 | Single config |
Cross-source percentile spread 3.6
Reasoning
#19High 0.9261.7
| Source | Percentile | Raw score | Config folding |
|---|---|---|---|
| Epoch AI | 56.3 | 0.9 | Median of 5 configs |
| LiveBench | 72.8 | 91.1 | Single config |
| LMArenatext-math | 53.9 | 1472.6 | Median of 2 configs |
| · text-hard | 57.8 | 1476.7 | Median of 2 configs |
Cross-source percentile spread 16.9
Value
#19High 0.9342.0
Aliases & configurations
- gpt-5-4
- gpt-5-4-2026-03-05
- gpt-5-4-high· high
- gpt-5-4-high-codex-harness· high / codex harness
- gpt-5-4-medium-codex-harness· medium / codex harness
- gpt-5-4-xhigh· xhigh