Long-lived GPT workhorse for coding, instruction following, and production apps
Specification
openai listing
Context
1M
Max output
33K
Released
2025-04-14
Knowledge cutoff
2024-04
Retires
2027-04-14
Open weights
no
Input
text, image, pdf
Output
text
Price
US dollars per million tokens · openai list price
Input
$2.00
Output
$8.00
Cache read
$0.50
Cache write
$2.00
Batch and priority prices
US dollars per million tokens · per listing
| Price | Listing | $/M |
|---|---|---|
| Batch input | azure | $1.00 |
| Batch input | azure | $1.10 |
| Batch output | azure | $4.00 |
| Batch output | azure | $4.40 |
| Priority input | openai | $3.50 |
| Priority output | openai | $14.00 |
Batch input is 50% below the interactive input price at the same listing.
Quality
10 benchmarks · 3 sources
Composite
41st
percentile of 334 scored models
Rank
196 / 334
±0.04 sd
Evidence
10 × 3
3 or more benchmarks, from 2 or more sources
Effort range
9.0
points between effort settings
coding34th2/3 bench
reasoning48th2/4 bench
math36th5/7 bench
knowledge28th1/1 bench
Every score
one row per source and configuration — nothing averaged away
| Benchmark | Score | Configuration | Source | Run |
|---|---|---|---|---|
| MATH level 5 | 83.0 ±0.8 | default | Epoch AI | 2025-04-14 |
| Aider polyglot | 52.4 | default · aider diff | Aider | 2025-04-14 |
| GPQA diamond | 66.9 ±2.8 | default | Epoch AI | 2025-04-14 |
| OTIS Mock AIME 2024-2025 | 38.3 ±6.3 | default | Epoch AI | 2025-04-14 |
| SimpleQA Verified | 31.1 ±1.5 | default | Epoch AI | 2026-08-31 |
| FrontierMath-Tier-4-2025-07-01 | 0.0 ±0.0 | default · 2 runs | Epoch AI | 2025-07-01 |
| SWE-Bench verified | 48.5 ±2.3 | default | Epoch AI | 2026-02-08 |
| ↳ | 39.6 | default · mini-SWE-agent | SWE-bench | — |
| Sources and settings disagree by 8.9 points. Both figures stand. | ||||
| Chess Puzzles | 6.0 ±2.4 | default | Epoch AI | 2026-08-07 |
| FrontierMath-v1 | 2.8 ±0.0 | default · 2 runs | Epoch AI | 2025-04-14 |
| FrontierMath-Tiers-1-3-v2 | 6.0 ±1.4 | default | Epoch AI | 2026-08-27 |
The leading figure for a benchmark is the median across its configurations, so one heroic high-effort run cannot set the number.
Available from 29 hosts
| Host | In $/M | Out $/M | Cache rd | Cache wr | Context | Output | Retires |
|---|---|---|---|---|---|---|---|
| openaivendor | $2.00 | $8.00 | $0.50 | — | 1M | 33K | — |
| poe | $1.80 | $7.20 | $0.45 | — | 1M | 33K | — |
| 302ai | $2.00 | $8.00 | — | — | 1M | 33K | — |
| abacus | $2.00 | $8.00 | $0.50 | — | 1M | 33K | — |
| azure | $2.00 $2.20 | $8.00 $8.80 | $0.50 $0.55 | — | 1M | 33K | 2027-04-14 |
| azure-cognitive-services | $2.00 | $8.00 | $0.50 | — | 1M | 33K | — |
| cloudflare-ai-gateway | $2.00 | $8.00 | $0.50 | — | 1M | 33K | — |
| edenai | $2.00 | $8.00 | $0.50 | — | 1M | 33K | — |
| fastrouter | $2.00 | $8.00 | $0.50 | — | 1M | 33K | — |
| helicone | $2.00 | $8.00 | $0.50 | — | 1M | 33K | — |
| impossibl | $2.00 | $8.00 | $0.50 | — | 1M | 33K | — |
| kilo | $2.00 | $8.00 | $0.50 | — | 1M | 33K | — |
| llmgateway | $2.00 | $8.00 | $0.50 | — | 1M | 33K | — |
| llmgateway-providers | $2.00 | $8.00 | $0.50 | — | 1M | 33K | — |
| merge-gateway | $2.00 | $8.00 | $0.50 | — | 1M | 33K | — |
| nano-gpt | $2.00 | $8.00 | $0.50 | — | 1M | 33K | — |
| nearai | $2.00 | $8.00 | $0.50 | — | 1M | 33K | — |
| ofox | $2.00 | $8.00 | $0.50 | — | 1M | 33K | — |
| openrouter | $2.00 | $8.00 | $0.50 | — | 1M | 33K | — |
| orcarouter | $2.00 | $8.00 | $0.50 | — | 1M | 33K | — |
| pioneer | $2.00 | $8.00 | $1.00 | $2.00 | 1M | 33K | — |
| replicate · replicate | $2.00 | $8.00 | — | — | — | — | — |
| sap-ai-core | $2.00 | $8.00 | $0.32 | — | 1M | 33K | — |
| vercel | $2.00 | $8.00 | $0.50 | — | 1M | 33K | — |
| vercel_ai_gateway · vercel-ai-gateway | $2.00 | $8.00 | $0.50 | free | 1M | 33K | — |
| cortecs | $2.192 | $8.769 | $0.546 | — | 1M | 33K | — |
| anyapi | — | — | — | — | 1M | 33K | — |
| github_copilot · github-copilot | — | — | — | — | 128K | 16K | — |
| model-oracle-ai | — | — | — | — | 1M | 33K | — |
More from OpenAI
most-hosted first